University of Central Florida STARS

Electronic Theses and Dissertations, 2004-2019

2008

Chromatographic And Mass Spectral Analyses Of Oligosaccharides And Indigo Dye Extracted From Cotton Textiles With Manova And Ano

Jessica Frisch University of Central Florida

Part of the Chemistry Commons, and the Forensic Science and Technology Commons Find similar works at: https://stars.library.ucf.edu/etd University of Central Florida Libraries http://library.ucf.edu

This Masters Thesis (Open Access) is brought to you for free and open access by STARS. It has been accepted for inclusion in Electronic Theses and Dissertations, 2004-2019 by an authorized administrator of STARS. For more information, please contact [email protected].

STARS Citation Frisch, Jessica, "Chromatographic And Mass Spectral Analyses Of Oligosaccharides And Indigo Dye Extracted From Cotton Textiles With Manova And Ano" (2008). Electronic Theses and Dissertations, 2004-2019. 3625. https://stars.library.ucf.edu/etd/3625

CHROMATOGRAPHIC AND MASS SPECTRAL ANALYSES OF OLIGOSACCHARIDES AND INDIGO DYE EXTRACTED FROM COTTON TEXTILES WITH MANOVA AND ANOVA STATISTICAL DATA ANALYSES

by

JESSICA LYNNE FRISCH B.S. University of Central Florida, 2005

A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in the Department of Chemistry in the College of Sciences at the University of Central Florida Orlando, Florida

Spring Term 2008

© 2008 Jessica Lynne Frisch

ii ABSTRACT

Research was conducted on thirteen 100% cotton denim samples using an acid wash, established

by Murray, to extract oligosaccharides from the cellulosic material. The oligosaccharide ion

+ + + groups ([M+H] , [M+NH4] , and [M-OH] ) for molecules with degrees of polymerization

between two and seven (DP2-DP7) were analyzed using liquid chromatography coupled to mass spectrometry with an electrospray ionization interface (LC-ESI-MS). The results were compared using the least-squares means in a Multivariate ANOVA (MANOVA) test followed by

Univariate ANOVA and Tukey HSD tests and demonstrated that the method could correctly determine that two samples were statistically different 85.9% of the time when analyzing the amount (ng) of each of the oligosaccharide ion groups separately, and 82.0% when analyzing the total moles of monosaccharide units released. A dye extraction was performed on the denim materials and the extract analyzed using gas chromatography coupled with mass spectrometry

(GC-MS). Indigo dye was present in all of the denim samples except one. When these results were combined with the two oligosaccharide statistical analyses, the discriminating power was increased to 88.5% and 85.9%, respectively. Additional cellulosic materials were also investigated including four white 100% cotton t-shirts as well as five raw cotton samples grown in Tajikistan, Uzbekistan, Egypt, Iran, and Benin West Africa. The analytical methodology gave results for the white cotton t-shirts and raw cotton samples that were inconsistent with those obtained from the denim samples.

iii

This thesis is dedicated to the memory of my Dad.

David Mark Frisch May 23, 1957 – September 2, 2005

He will always be loved and shall forever live on in my heart.

iv ACKNOWLEDGMENTS

I would like to express my sincere gratitude to: my mom, Cheryl Frisch – for her prayers; Uncle

Bill and Aunt Nora – for always being there for me; my step-mom, Kathy – for supporting me

and my education; my youngest sister, Clara, and my brother, Robert – for always making me

smile and inspiring me to be the kind of individual they can look up to; my supportive friends

Alisha and Bryan Yabczanka, Clare Kidwell, Nathan Costa, Kyle and Erin Parker, LeAnn and

Ryan Lett, Diane Kodera, and my boyfriend Chris Daiello – for their continuous encouragement

and friendship; my co-workers past and present at NCFS, Mary Williams and Katie Steele –for

demonstrating how to successfully defend a thesis; Doug Clark – for putting up with questions,

despite how ridiculous; “Miss Judy” Stout – if it was not for her chocolate candy stash, I do not know how I would have made it through all the “Mondays”; April Marrone – for her trouble- shooting help with the instrument and for making 242 my new “favorite” number; my co-worker, roommate and friend, Erin McIntee – for her help with what seemed like endless sample extractions and having the patience to hear me talk about my research/thesis endlessly; Dr.

Candice Bridge – for inspiring me to go for my Ph.D. I would like to thank Dr. Fookes and Dr.

Hampton for serving on my thesis committee and helping me when I needed it, and Dr. Ni for all

of his help with statistics. Finally, I would especially like to thank Dr. Sigman for all of his help,

support, and for challenging me. I could not have asked for a better adviser.

“A mind, once stretched by a new idea, never returns to its original dimensions." Oliver Wendell Holmes

v TABLE OF CONTENTS

LIST OF FIGURES ...... ix

LIST OF TABLES...... xi

CHAPTER 1: INTRODUCTION...... 1

CHAPTER 2: BACKGROUND INFORMATION AND LITERATURE REVIEW ...... 4

2.1 Cotton...... 4

2.1.1 Cotton Introduction...... 4

2.1.2 Cotton Structure...... 7

2.1.3 Classifying and Manufacturing Cotton for Textiles ...... 13

2.1.4 Cotton Fiber Frequency and Discrimination...... 21

2.2 Oligosaccharide Analyses and Differentiation of Cellulosic Materials...... 25

2.2.1 Work by Allen K. Murray...... 26

2.2.2 Additional Methods of Analyses ...... 30

2.3 Instrumentation ...... 38

2.3.1 LC-ESI-MS...... 38

2.3.2 GC-MS...... 54

2.4 Statistical Methods of Analyses...... 57

2.4.1 One-way Analysis of Variance ...... 57

2.4.2 Multivariate Analysis of Variance...... 61

CHAPTER 3: OLIGOSACCHARIDE ANALYSIS ...... 64

3.1 Instrument Parameters ...... 64

3.2 Oligosaccharide Extraction and Isolation from Cellulosic Material...... 68

vi 3.2.1 Oligosaccharide Acid Extraction ...... 69

3.2.2 Isolation of Oligosaccharides...... 69

3.3 Standards...... 71

3.3.1 Sample Collection and Preparation...... 71

3.3.2 Analyses and Results ...... 71

3.4 Denim...... 78

3.4.1 Sample Collection and Preparation...... 78

3.4.2 Analyses and Results ...... 81

3.5 White Cotton T-Shirts...... 99

3.5.1 Sample Collection and Preparation...... 99

3.5.2 Analyses and Results ...... 100

3.6 Raw Cotton ...... 103

3.6.1 Sample Collection and Preparation...... 103

3.6.2 Analyses and Results ...... 104

CHAPTER 4: DYE ANALYSIS ...... 107

4.1 Instrumentation Parameters ...... 107

4.2 Standards...... 107

4.2.1 Sample Collection and Preparation...... 107

4.2.2 Analyses and Results ...... 108

CHAPTER 5: DISCUSSION...... 112

5.1 Current Research...... 112

5.2 Future Research ...... 115

vii APPENDIX A: SAS CODE FOR DENIM SAMPLES USING AMOUNT OF

OLIGOSACCHARIDES (NANOGRAMS) FOR DIFFERENT ION GROUPS...... 117

APPENDIX B: SAS CODE FOR DENIM SAMPLES USING TOTAL MOLES OF

MONOSACCHARIDES...... 130

LIST OF REFERENCES...... 143

viii LIST OF FIGURES

Figure 1: Oligosaccharide Structure for Cellulose Including Cellobiose Repeating Unit...... 9

Figure 2: Morphology of Secondary Wall...... 12

Figure 3: Diagram of 3/1 Right and Left Hand Twills ...... 15

Figure 4: Process Flow for Warp Yarn in Denim Manufacturing ...... 17

Figure 5: Structures for Indigo and Indigo Derivatives...... 18

Figure 6: LC-API-MS Schematic ...... 39

Figure 7: ESI Spray Schematic...... 43

Figure 8: Finnigan Corporation LCQ ESI-QIT-MS ...... 46

Figure 9: MS Schematic...... 47

Figure 10: QIT Schematic...... 49

Figure 11: QIT Stability Diagram...... 52

Figure 12: GC Schematic...... 56

Figure 13: Fragmentation of Oligosaccharide-Ammonium Adducts ...... 67

Figure 14: Extracted Ion Chromatograms for 10 ppm DP1-7 Standard...... 73

Figure 15: 3-D Map of Chromatogram for DP1-7 Standard...... 74

Figure 16: Extracted Ion Chromatogram of DP7 Standard Following Rotary Evaporation...... 76

Figure 17: Extracted Ion Chromatogram of DP7 Standard Following Oligosaccharide Extraction

Process ...... 78

Figure 18: Extracted Ion Chromatograms for Each Denim Sample...... 82

Figure 19: 3-D Map of Chromatograms Denim Samples S6 and S2...... 83

Figure 20: Sample Standard Calibration Curves for Each Ion Group ...... 85

ix Figure 21: 3-D Map of Chromatograms for a DP1-7 Standard and Denim Sample S5 ...... 89

Figure 22: Tukey HSD Results of Pairwise Comparisons...... 93

Figure 23: Tukey HSD Results of Pairwise Comparisons Using Total Moles of

Monosaccharides...... 97

Figure 24: Extracted Ion Chromatograms for White Cotton T-Shirt Samples ...... 101

Figure 25: 3-D Map of Chromatograms for a White Cotton T-Shirt Sample J, Extractions 1 and

2...... 102

Figure 26: Extracted Ion Chromatograms for Raw Cotton Samples ...... 105

Figure 27: 3-D Map of Chromatograms for a DP1-7 Standard and Raw Cotton Sample R1.... 106

Figure 28: Extracted Ion Chromatogram and Spectrum for Indigo Standard...... 109

Figure 29: Extracted Ion Chromatogram and Spectrum for Denim Sample S3a ...... 110

Figure 30: Isatin ...... 110

Figure 31: Comparison of Tukey HSD Results for Both Statistical Methods...... 113

Figure 32: Discrimination Matrix Based on Different Ion Groups and the Presence of Indigo

Dye...... 114

Figure 33: Discrimination Matrix Based on the Total Moles of Monosaccharides and the

Presence of Indigo Dye...... 115

x LIST OF TABLES

Table 1: Types of Errors ...... 2

Table 2: Solvents and Additives Suitable for API-MS...... 40

Table 3: Mass-to-Charge Ratios of the Ions/Adducts Corresponding to Oligosaccharides with

Degrees of Polymerization 1-7 ...... 68

Table 4: 100% Cotton Denim Fabric Samples Purchased from Craft Sections ...... 80

Table 5: Equations for Denim Sample Calibration Curves and Respective R2 Values...... 86

Table 6: Number of Extractions and Observations within the Linear Dynamic Range of the

Calibration Curve for the Denim Samples (Including Sample Repeats S9 and S15)...... 88

Table 7: Results of Individual Univariate Analyses on the Ion Grouping Criterion Values ...... 92

Table 8: Summary of Tukey HSD Results and Calculated Percent Discrimination...... 94

Table 9: Amount (ng) of Each Ion Group Detected and the Total Moles of Monosaccharides

Extracted for Sample Size Reduction Test ...... 98

Table 10: White 100% Cotton T-Shirt Samples ...... 100

Table 11: HVI Data for ICA Ltd. Samples...... 103

xi CHAPTER 1: INTRODUCTION

The majority of fibers in the transferable fiber population are cotton. These fibers lack any

significant evidentiary value, because they are so common. It is the goal of this research to

investigate the possibility of increasing the evidentiary value of cotton fibers, specifically denim

fibers.

“Cotton (Gossypium spp.) is the leading textile fiber in the world and one of the most important

oilseed crops” [1]. Cotton from the cotton boll of a plant is processed to be manufactured into

textiles such as t-shirts, denim jeans, bathroom towels, etc. Examining the fabric closely would

reveal hundreds to thousands of individual cotton fibers. Each cotton fiber is a single elongated cell that has dried to leave only the cell wall. The cell wall is composed of cellulose, which is polysaccharide composed of repeating cellobiose units. Cellobiose is a disaccharide composed of two glucose molecules connected via a β-1,4-glucosidic linkage where each glucose molecule is rotated approximately 180° with respect to the previous. The term “oligosaccharides” is used when a polysaccharide consists of only a few monomer units, and the simplest oligosaccharide form is a disaccharide composed of two monosaccharides [2]. Glycan is another name for

“polysaccharide” [2]. The degree of polymerization (DP) is found using Equation 1:

MW DP = polymer (1) MWrepeating unit

1 For example, a molecule that is considered DP6 consists of six glucose molecules connected to one another, and a DP1 oligosaccharide is actually a monosaccharide consisting of only a single glucose molecule. In this research DP2, DP3, DP4, DP5, DP6, and DP7 oligosaccharides will be extracted from denim material and the oligosaccharide content (ng oligosaccharides / 60.0 mg fabric) analyzed as a potential means of discriminating between different denim fabrics.

When questioned and known fibers are compared, it is important for the “association error” to be

very small. Association errors arise due to the manufacturing process which produces mass

amounts of textiles, thereby decreasing the uniqueness of the fibers. When comparing a

questioned and known sample, there are three possible outcomes: the fibers are consistent with

one another, are not consistent with one another, or the results are inconclusive. The first two

possibilities may be correct or incorrect. If the results that two fibers are not consistent with one

another is incorrect, then a type I error has occurred. If the results that two fibers are consistent

with one another in incorrect, than a type II error has occurred [3] (Refer to Table 1).

Table 1: Types of Errors Conclusion Fiber Did Originate Fiber Did Not Originate from Known Sample from Known Sample Consistent Correct (P = 1 - α) Type II error (P = β) Not Consistent Type I error (P = α ) Correct (P = 1 - β)

Type II errors are the most detrimental to a forensic scientist, because they could lead to the

conviction of an innocent individual. The actual probability of a type II error in fiber analysis is

not known, but is affected by whether or not the fibers and materials are mass produced, their

frequency of occurrence at a crime scene (i.e. matching only one or several fiber(s) from a crime

2 scene to a suspected textile material), whether or not the fiber contains a characteristic attribute

(i.e. it has been burned or there is paint on it), fibers that are found in unexpected locations, and

whether or not a two-way transfer has occurred [3]. Additional research that may be conducted

in order to discriminate between fabrics would be the analysis of the fiber dyes, finishing agents, exposure/wear studies, and laundry detergents [3].

3 CHAPTER 2: BACKGROUND INFORMATION AND LITERATURE REVIEW

2.1 Cotton

2.1.1 Cotton Introduction

The cotton plant is of the genus Gossypium, and there are four principle domesticated species

that are used commercially, which include hirsutum, barbadense, aboreum, and herbaceum.

Each species has several different varieties, some of which have been developed through breeding programs to improve the individual cotton characteristics such as strength, fiber length, and resistance to diseases. While cotton is generally grown for fiber, the cottonseeds are also a food crop used for human consumption, animal feed, and vegetable oil, and the linters are used for chemical cellulose and batting [4].

Cotton is grown in approximately 80 countries. Between the years 2004-2005, 12,350 acres of cotton were cultivated in 59 countries, raw cotton was exported from 57 countries, and cotton textiles were exported from 65 countries. Annually, approximately 90,000,000-150,000,000 bales are produced world-wide, with 18,000,000-20,000,000 US bale equivalents produced in the

US. In 2004, about 40% of the fiber consumed worldwide was cotton. The world production of raw cotton fibers consists of 90% Upland cottons. American Upland cottons fall in the hirsutum species, and were developed in the United States from cottons that were native to Central

America and Mexico. The fibers are between 22-36 mm in length (26-30 mm for lint fibers

4 grown in the U.S.), with micronaire values ranging from 3.8-5.0. Upland cotton fibers are used

for home furnishings, apparel, and industrial products [4].

G. barbadense is also known as extra-long-staple (ELS) cotton, because it has the longest staple length (33-36 mm) of the cotton species. ELS cotton originated from South America and makes up 8% of the world cotton fiber produced. ELS cotton is used for luxury fabrics, sewing thread, and quality apparel. Some common varieties of ELS cotton are Egyptian, American-Egyptian,

Sea Island, and Pima. The Pima variety was created from a complex cross between Egyptian and

American Upland cottons, and it is primarily grown in the western United States [4].

Upland cotton produces a white fiber while Pima produces a creamy yellow fiber. Varieties of

naturally pigmented cottons also exist and are spontaneous mutants of white fiber producing

plants. Cotton may be naturally colored such as light tan, cinnamon, champagne, and green [5].

The color of the brown and red-brown cottons results from vacuolar tannin material found in the lumen, and the green cotton color results from suberin sandwiched between the lamellae of the secondary wall microfibrils [4]. The green cotton fibers also have higher wax content than the white and brown fibers.

Both G. hirsutum and G. barbadense are tetraploids; however, G. aboreum and G. herbaceum are diploids. The latter two species of cotton are known as “Desi” cottons and range from 9.5-19 mm in length, and is classified as “coarse” with a micronaire value greater than 6.0. These species are not as commercially important as the previous species, and are grown in Pakistan and

5 India (G. arboretum is also grown commercially in Thailand, Bangladesh, Vietnam, and Burma)

[4].

Transgenic or biotech cottons are produced by research groups using recombinant DNA

technology. Biotech cotton Bacillus thuringiensis (Bt) was created in 1996 and resulted in

cotton being one of the lead genetically engineered crops. The biotech cotton is more resistant to

insects and herbicide tolerance. Biotech cotton accounted for 35% of the cotton produced in the

world in 2004-2005 [4].

Cotton is harvested either mechanically or by hand. Greater than 99% of cotton crops in the

United States and Australia are harvested mechanically. Harvested cotton consists of cotton fibers attached to the cotton seed and foreign plant matter impurities. Once harvested, the cotton is taken to the ginning plant in trailers or modules. Modules are used for storing cotton in the

United States [4].

Ginning removes the plant matter and seed from the cotton fibers and also involves adjusting

moisture content and packaging. Upland cotton is ginned on saw gins; whereas, ELS cotton is

ginned on roller gins. The raw cotton is then formed into densely packed bales. The

International Organization for Standardization (ISO) 8115 states that a densely packed raw

cotton bale should have the following dimensions: length 1400 mm (55 in), width 530 mm (21

in), and height 700-900 mm (27.5-35.4 in), with a density between 360-450 kg/m3 (22.5-28.1

lb/ft3). On average, a US bale has the dimensions 1400 mm x 533 mm x 736 mm with a density

of about 448 kg/m3, and the bale weighs approximately 490 lbs (222 kg) with a single pound

6 containing 100 million or more individual fibers. The average weight of a bale in the rest of the

world ranges from 375-515 lbs (170-233 kg), depending on the country where the bale is

produced [4]. Samples of the bales are examined and classified based on the U.S. Department of

Agriculture (USDA) classing office, based on their fiber quality and physical properties which would effect the quality of the finished textile product and the efficiency of the manufacturing

process [4] (See section 2.1.3 Classifying and Manufacturing Cotton for Textiles). The bales of

cotton are then marketed.

2.1.2 Cotton Structure

Cotton fibers develop in four stages: initiation, elongation, secondary wall formation, and maturation [6]. One to two weeks after being planted, seedlings emerge from the soil, and 5-6

weeks later flower buds form. The cotton boll is considered the fruit of the cotton plant and consists of a seed pod which reveals cotton fibers when it flowers. Cotton fibers “originate as outgrowths from a subset of the epidermal cells of an unfertilized seed (ovale) when the flower first opens (anthesis)” [7]. A single boll contains three to five segmented compartments which together hold 20 to 30 ovules. The flower itself blooms for a single day, starting with white petals which change to bright pink by the end of the day and fall off the boll (carpel) by the end of the second day. The blossoms (white for Upland, creamy yellow for Pima, or dark-yellow) appear three to four weeks later. It takes between 40-80 days for the blooms to form into an open bolls [4]. Two types of fibers are produced: long lint fibers that are used for producing textiles and short linters (or fuzz fibers) which are used for paper, manufacturing plastics and rayons, or used as batting and padding in bedding, upholstered furniture and automotives. The fuzz fibers are not produced in G. barbadense cotton. [4]. Most fuzz fibers are produced near the

7 tip of the cotton seed, while the base of the seed produces mostly lint fibers. There are four

characteristics that are used to differentiate lint from linters: length, pigmentation, strength of

adherence to seed, and chemical and physical properties. The length of the average lint fiber is

25.4 mm (1 in.), where first-cut linters are between 12-15 mm (0.5 in.). Linters are usually highly colored or have light brown pigmentation and adhere more tightly to the seed. Linters are also coarser than lint fibers and lack the convolutions that the lint fibers possess. Linter fibers do not have as much of a tapered tip as lint fibers and hardly have any lumen compared to lint fibers. Lint fibers generally have half the width of linter fibers, and the primary cell wall growth for lint fibers starts on the day of anthesis, while linter fibers “initiate elongation” approximately four days later [4].

The cotton fiber is a unique plant cell that usually does not divide or store starch [8, 9]. It is

composed of approximately 94% cellulose, 1% protein, 1% pectin substances, as well as wax,

ash, and pigments [10]. Fiber cell elongation begins on the day of anthesis in an air space

enclosed by three to four carpels of the fruit [6]. Elongation continues for 15-30 days with the

formation of the primary wall containing one nanogram of cellulose per millimeter of fiber

length. The primary wall also contains hemicelluloses and pectin and is surrounded by a cuticle

composed of pectin and wax [8, 9]. After twenty to thirty days, or possibly during the overlap of

five to ten, the more dense secondary wall begins to form at a rate of 130 ng of cellulose per

millimeter of fiber [6, 10]. The twisted “ribbon-like structure” of the cotton fiber occurs between

45-60 days post anthesis (DPA) [6], when the fruit capsule opens and the fiber cell dies resulting in a dried, hollow cell wall which collapses on itself [4, 11]. The twists of the fiber alternate directions and are called “reversals”, which are related to the orientation of the microfibrils in the

8 secondary cell wall [6]. These reversals aid in the spinning of the fibers into yarns [4].

According to Shenouda, for native cotton from different origins, there are statistical differences

between the distances of the surface ridges [11].

Cellulose “is the most abundant organic compound on the planet” [12] and is composed of β-1,4- linked D-glucose monomers with cellobiose repeating units. Cellulose is synthesized by linear polymerization of glucose residues in which every other glucose residue in the β-1,4-glucosidic linkage is rotated approximately 180° with respect to the previous glucose residue [2, 13, 14].

This results in the repeating unit for this polymer as being the cellobiose disaccharide [13], and a cellulose molecule may be 500-15,000 glucose units in length [15] (Refer to Figure 1). (Note: A

glucose residue is defined as the portion of the glucose molecule remaining after the monomers are covalently linked following the removal of water [2].)

Figure 1: Oligosaccharide Structure for Cellulose Including Cellobiose Repeating Unit

Natural cellulose has a cellulose I crystal structure; whereas, regenerated cellulose is composed

of the cellulose II crystal structure, and the structures III and IV are only obtained under special

9 conditions [16]. The natural cellulose structure is composed of a unit cell made-up of “two identical and parallel chains with the C6 hydroxymethyl group in the trans gauche position”

which allows the formation of two intramolecular hydrogen bonds (02’-H--06 and 03-H--05’)

that are “parallel to the glycosidic linkage per anhydroglucopyranose unit” [16], forming the

cellulose I crystal in which the glucan chains are aligned parallel to one another and packed side

by side to form microfibrils [13, 14]. The chain length may range from 2,000 to 20,000 glucose

residues, depending on the organism, and “cellulose I may exist in two different allomorphs,

called Iα and Iβ, that may be distinguished by 13C-NMR” [13]. Cotton is stated as consisting primarily of Iβ [14] and 20% cellulose Iα [15].

There is controversy as to whether or not microfibrils are actually composed of elementary

fibrils, which are the smallest structural units of cellulose [10]. The theory of elementary fibrils

was first introduced by Fey-Wyssling in 1954, who claimed that there are an average of four

elementary fibrils (30-60 Å in diameter) to one microfibril (approximately 250 Å but may be between 100 and 500 Å) which then make up a fibril (2000 Å diameter) [10, 11]. The microfibrils also align parallel to one another to form larger bundles called fibrils (also known as

macrofibrils) that range from 90-120 nm in diameter [12, 13]. The packing of elementary fibrils

in microfibrils varies for cellulose from different sources [11]. The diameter of the microfibrils

that make-up the primary wall are approximately 12-25 glucan chains thick, while the secondary

wall microfibrils are larger with a width between 3-4 nm [6]. The elementary fibers are believed

to have a width of approximately 3.5 nm [10]. This cellulose structure makes up 95-97% of the

plant cell wall of the cotton fiber [12].

10 It is difficult to determine the molecular weight of individual cellulose molecules since most

solubilization techniques are so harsh they degrade the individual molecules; however, nitration

of cellulose using a mixture of phosphorous, phosphoric acid, and nitric acid breaks the

glycosidic bonds [15]. The secondary wall consists of 14,000 to 15,000 (12,000 to 15,000 [6]) fairly uniform degrees of polymerization [15]. Cellulose in the primary wall has a lower molecular weight with degrees of polymerization that may be below 500, and portions of the primary wall may range between 2000-4000 DP, as well as intermediate fractions [15]. Other sources claim that cotton cellulose varies from 5,000 to 12,000 DP for untreated cotton [10].

The elongation process involves the fibers’ primary cell wall covered by the cuticle growing

longitudinally for 21-35 days post-anthesis (DPA) until they reach their final length (1-6 cm),

which is controlled by the different cotton genotypes as well as environmental conditions [4].

The cuticle of the cotton fiber is typically 10-20 nm thick [6]. The primary cell wall is composed

of randomly oriented carbohydrates and protein macromolecules, and the sugar composition is

observed to change as the fiber develops [4]. Between 15 and 19 DPA, the secondary wall of the lint fiber cells is produced which consists of nearly pure cellulose [β-(1,4)-D-glucan] in a highly

crystalline structure. (Note: some callose, β-(1,3)-D-glucan is also formed between the

secondary wall and the plasmalemma.) This process continues for 30-45 days until the

secondary wall is greater than 90% cellulose with a secondary cell wall thickness between 2-6

µm. The secondary wall is produced in layers of microfibrils that are deposited in a helical

fashion with a gradual change in angle in relation to the fiber axis that are directly related to the

strength of the fiber. Occasionally the angle will change so that the microfibrils are aligned in

11 the opposite direction, known as a reversal, and it is here that the cellulose linkages are most

easily broken [4].

The primary cell wall fibrils are aligned in a crisscross pattern [11]. The secondary wall is more

dense with more microfibrils and is divided into three layers: s1, s2, and s3 [10] (See Figure 2).

Figure 2: Morphology of Secondary Wall

In all the layers the cellulose microfibrils are arranged in a helical pattern around the fiber axis;

however, the s2 layer helices are at a steeper angle than the s1 and s3 layers. Reversals in the helical direction occur in both the s1 and s2 layers so the directions are always opposite of one

another [10]. “[A]fter the cell completes its longitudinal growth” the s1 layer is deposited [11],

and is known as the “winding layer”, with microfibrils aligned at a 45 to 55 degree angle to the

12 fiber axis [6]. Also known as the “transitory lamellae”, the s2 layer makes up most of the

secondary wall and is composed of fibrils that are arranged in a helical pattern around the lumen at an acute angle of 20-30 degrees to the fiber axis [11], decreasing to -15 degrees as the fiber reaches maturity [6]. Reversals also occur in this layer along the length of the fiber, also known as spiral angles [11]. An s3 layer is also believed to exist which connects the secondary wall to the hollow space inside of the collapsed fiber, known as the lumen [11].

2.1.3 Classifying and Manufacturing Cotton for Textiles

Cotton bales are classified based on many properties that affect the quality of the cotton. When

classifying Upland and Pima cotton, the United States Department of Agriculture (USDA) using

High Volume Instrument (HVI) systems, patented by Uster Technologies. Both species of

cotton are classified based on leaf grade (amount of leaf material present with the cotton fibers),

extraneous matter (any matter other than fiber and leaf), instrument measurements for color grade (unofficial for Pima cotton), fiber length, micronaire, strength, length uniformity index,

color Rd, color +b, and trash percent areas. The Upland color grades are determined using the

HVI instrument and are two-digit numbers; whereas, the Pima cotton color is graded by classers

instead of by the HVI instrument. The fiber length is denoted in both thirty-two-secondths of an

inch (columns 34-35) and in one hundredths of an inch. Micronaire of the cotton is a measure of

the cotton’s fineness, which is highly correlated to maturity, within a cotton variety. Micronaire

is measured by the resistance to air flow per unit mass. The strength of the fiber is measured in

grams per tex, which represents the force in grams required to break a bundle of fibers that is one

tex in size. The tex unit is defined as the weight in grams of one thousand meters of fiber. The

13 classification data is summarized in the Universal Classification Data Format which is attached to the incoming bale identification tag in order to provide the data to other organizations/consumers/etc [17].

Cotton textiles are composed of two widely grown species of cotton: Gossypium hirsutum

(Upland cotton) and Gossypium barbadense (Pima cotton) [18]. Ninety percent of cotton textiles are composed of Upland cotton and 8% of Pima cotton. The hirsutum species is used for 99.9% of denim textiles. Many of the varieties of hirsutum species that are grown commercially are more likely to share some genetic background. Roughly 50 bales of cotton (one bale is approximately 500 pounds) from all over the world, and possibly representing multiple varieties of cotton, are used to generate yarns in a textile mill. Each bale has a certain distribution of fiber properties, which are described by the HVI test values [18]. The bales are constantly blended together when creating the yarn.

2.1.3.1 Denim

Denim fabric consists of two sets of yarns woven perpendicular to one another such that the warped yarn is interlaced by picks or filling yarns [19]. By varying the number of warped vs. filling yarns, the fabric weight, tightness, cover, drape, hand, tensile, strength, and tear strength may differ, allowing the fabric to be constructed for a specific use or item of clothing. Unlike conventionally woven fabrics, the warped yarn produced for denim fabric is prepared by specific processing steps that are unique to denim fabric. Classic bottom weight 14.5 ounce denim is composed of 60-64 warp yarns per inch and 38-42 filling yarns per inch. Denim fabric used for

14 blouses, tops, shirts, and top of bed fabrics is usually between 3.5 and 8.0 ounces per square yard, and between 8.0 and 16.5 ounces per square yard for trousers, jeans, jackets, and upholstery [19].

Denim is woven in a 3/1 twill, 2/1 twill, 3/1 broken twill, or 2/2 broken twill pattern. When the pattern is 3/1 twill, it means that each warp yarn is going over three filling yarns then under one filling yarn. When this process is repeated in the next row, the pattern is moved up one filling yarn. As this process continues, a right-hand twill weave is produced; however, a left-hand twill weave may also be produced (Refer to Figure 3) [19].

Figure 3: Diagram of 3/1 Right and Left Hand Twills

15 Broken twill weaves do not form a straight line. When the manufacturer desires a more pronounced twill line, the warp yarn used should have a twist that is opposite the twill direction in the fabric, which will also make the fabric softer. When an S-twisted yarn is woven into right- hand twill, then the twill line would be more pronounced, and a Z-twisted yarn would produce a less-pronounced twill line. Ring-spun yarns may produce both S- and Z-twisted yarns; however, open-end yarns only produce Z-twist yarns. A Z-twist yarn is formed when the spindle is rotated clockwise, and a S-twist yarn is produced when the spindle is rotated counter-clockwise [19].

Prior to the 1970s, all denim yarns were ring spun; however, today ring spun, open-end spun, and a combination of the two are used (i.e. ring/ring, OE/OE, or ring/OE) where the first mentioned is the warped yarn and the second is the filling yarn (i.e. warp/filling) [19]. While ring-spun fabric characteristics are favorable, the combination of weaving with open-end yarns reduces the cost of the fabric.

Specialty yarns are constructed for different types of denim materials. For example, elastic denim yarns are commonly used in “stretchy” jeans. These yarns are formed when the cotton fibers are twisted around a spandex filament. “Antiqued” yarns and design patterns in the fabric may also be produced by varying the instrumentation used for weaving [19].

Unlike normal woven fabrics, denim is unique in that the warp yarns are specifically processed and dyed before being placed on the weaving machine; however, the filling yarns are fed into the weaving machine without any additional preparation. Figure 4 is a flow chart representing the steps which are necessary to manufacture denim fabric [19].

16

Figure 4: Process Flow for Warp Yarn in Denim Manufacturing

The warping process transfers multiple yarns from individual yarn packages onto a single

package assembly [19]. Beam warping collects the yarns in a sheet form so that the yarns lie

parallel to one another; whereas, in ball warping the yarns are condensed together into a rope

before being wound into a barrel. Both warping techniques wind the yarns onto spindles in what

is then called a creel. When beam warping is used, the yarns do not pass through the rope indigo

dye range, but instead will be left un-dyed or undergo slasher dyeing [19].

While natural un-dyed denim may be produced, denim is typically dyed with indigo (Color Index

No. 73000, Refer to Figure 5), which may also be mixed with small amounts of other dyes such as sulfur blue, sulfur black, hydron blue, and the indigo structure may be substituted with chlorine, bromine, or methyl groups on the basic structure (Refer to Figure 5) [20]. The sulfur

17 dyes, including hydron blue, are classified according to the chemical structure of their starting material, since little is know about the structure of the sulfur dyes [21].

Figure 5: Structures for Indigo and Indigo Derivatives

18 When indigo dye is in its normal form, it is vibrant blue in color and insoluble in water;

therefore, the dye will not adhere to the cotton fibers [19]. In order to dye the cotton yarns for denim production, the indigo dye must first be converted into its “leuco” form, which is water- soluble and yellow-green in color. The conversation of indigo dye to its leuco form involves a chemical reduction using reducing agents such as sodium hydroxide with sodium hydrosulfite.

Once in the leuco form, the dye is absorbed by the outer layers of the cotton yarns and is then oxidized back to the insoluble form as the dye comes into contact with the oxygen in the air (a process known as “skying”). This “ring dyeing” process traps the indigo dye in the outer layers of the yarn, leaving a white center. Since the dye is not readily soluble in the cotton, only a small amount of dye adheres to the yarns; therefore, in order to produce a denim fabric with a deeper shade of blue, the yarns must repeat the dyeing and drying processes (known as “dips”) several times [19].

If an even darker fabric is desired, a sulfur black dye may be applied to the yarn either before the

indigo dyeing process or after, which is known as sulfur bottom or sulfur top respectively [19].

Like indigo dye, the sulfur dye must be reduced to a water-soluble form; however, unlike indigo

dye, the sulfur dye is absorbed through the entire cotton fiber. In rope dyeing, the ball warps

may pass through scouring baths first, which consist of caustic, wetting agents, and detergents.

Scouring removes impurities typically found in the cotton fiber, such as dirt, ash, pectin, and

waxes. When these are removed, the cotton is more susceptible to uniform wetting and dyeing.

After scouring, the ropes are rinsed in water balls, and then sulfur dyes may be applied if a sulfur

bottom is desired. The ball warps undergo indigo dye dips, with skying after each dip, then are

19 rinsed with water before passing through a sulfur dye bath, if a sulfur top is desired. Finally, the yarn ropes pass through squeeze rolls, are dried, and coiled onto large tubs.

A sheet of yarn is dyed using a slasher dyeing range, and is particularly useful when dying lightweight denims. “In general, these machines require less floor space, enable smaller production runs, have a quicker turn over time, and are more flexible in their response to changes in the market” [19]. These machines may also be used for dyes other than indigo in order to dye cotton.

Beam dying is a technique that is capable of dying hundreds of individual yarns that are wound parallel to one another. These yarns are passed over a perforated core beam which is loaded onto a sealed cylindrical dye vessel. Dye liquor is pumped through the perforations in the beam and onto the yarn. The yarn is then washed, extracted, dyed, and continues to other beams in order to undergo slashing and weaving. This technique is generally used for dyes other than indigo and may be used for fabrics other than denim [19].

Following the dyeing of the warp yarns, the yarns are then aligned in a sheet form using a re- beaming process in order to undergo slashing. Slashing, also referred to as sizing, “encapsulates the yarn with a protective coating” which reduces yarn abrasions and hairiness, allows the weaving process to be more efficient, and reduces the amount of indigo dye that is rubbed off during weaving [19]. During the weaving process, the warp yarns on the loom beam are directed through the waving machine, where each yarn is fed through drop wires, heddles, and a reed.

The reed resembles a comb, with the distance between the spaces in the reed, known as dents,

20 ranging between 12 to 18 dents per inch for denim fabric. Filling yarn is fed from a location

outside of the weaving machine.

2.1.4 Cotton Fiber Frequency and Discrimination

Textile fibers fall under two main categories: natural and man-made. Natural fibers may be

further sub-classified as originating from animal, vegetable, or mineral, and man-made fibers to synthetic polymer and natural polymer [22]. Cotton fibers fall under the natural/vegetable classification, while regenerated cellulose falls under the natural polymer classification. Using polarized light microscopy, cotton fibers are easy to identify. Natural cotton fibers will “appear as a regular-to-irregular-twisted ribbon” which will not go completely into extinction when viewed under crossed polars [23].

The forensic science area of trace evidence is based on Locard’s Exchange Principle that “every

contact leaves a trace”. The more common the fiber, the less evidential value it has [22].

Common fiber types include white cotton, certain blue cotton fibers (usually dyed with natural or

synthetic indigo dyes), off-white cotton, and some non-dyed synthetic fibers. While this lack of

evidential value limits the ability to state a positive match between a fiber and its source, the

fibers should still be collected. If common fibers are present in a high concentration than what

would be expected, their value may become more significant [23].

According to Houck, population and target studies are used to help characterize the significance of trace evidence. A population study is “designed to evaluate the concentration of a randomly

21 sampled fiber population by dividing it into generic type/color combinations”, such as studying

textile fibers found on car or cinema seats. A target fiber study, usually case-related, is

“designed to establish how often a selected fiber may be expected to appear among a randomly

sampled fiber population” [23].

A fiber target study on textile cinema seats during the winter in Sydney, Australia, was

conducted by Cantrell, Roux, Maynard, and Robertson. Sixteen seats were analyzed and 3025

fibers were classified based on their generic class, color, and fluorescence properties, although

white and colorless fibers were excluded from the study. Eighty-four percent of the fibers found

were classified as natural fibers and cotton fibers made-up 70%. Fifteen percent of the fibers

were man-made with the largest percentage of those (51%) being classified as rayon. The most

common generic class and color combination were grey-black cotton fibers at 33% followed by

blue cotton at 30%, for an overall 63% of the total population. Only the grey-black cotton fibers

were analyzed for their fluorescent properties, since they were the fibers more commonly

encountered. Cantrell et al. also reported on a previous population study by Grieve and

Biermann where cotton was found to have a frequency of 75% in the spring and summer in

Germany [24]. In 2002, a study was conducted in Cambridgeshire, UK on the population of textile fibers in human hair. From the 26 volunteers, more than 12,000 fibers were collected and classified by type, color, and fiber length. Just over 72% of the fibers were classified as natural fibers with the majority being cotton fibers. The most common fibers encountered were black cottons followed by blue cottons [25]. Another population survey was conducted in the spring of

2003 in Sydney, Australia. This study analyzed colored fiber samples from eleven front- and top-loading washing machines, and over 12,000 fiber samples were classified based on generic

22 type, color, and length. Cotton fibers were found to be the most prevalent at just over 69%. The

most common generic class and color combination were black/grey fibers at 27% followed by

blue cotton at 20% and red cotton at just over 15% [26].

The majority of blue jeans consist of blue and white cotton fibers woven in a twill pattern.

Indigo is the most commonly used dye for blue jeans, although the shading of the dye may vary

along blue jean fabric. Since indigo is commonly used, it is often impossible to distinguish

between different blue jeans unless whiteners or brighteners have been added to the fabric. In a

case studied by Max Houck, he compared fibers found at a crime scene to denim jeans owned by

two individuals. The two individuals had the same brand of denim jeans. Houck was not able to

determine the source of the fibers [23].

Denim fibers are often excluded from fiber discrimination studies since they are already known

to have very little evidential value. Max Houck excluded both colorless and denim fibers from

his study on the inter-comparison of unrelated fiber evidence, and he stated “except for blue

denim or grey/black cotton fibers dyed with sulfur black 1, no fiber should be considered as

common” [27]. According to an article published in 1988 by Grieve, Dunlop, and Haddock,

“[f]ibers from blue denim cannot generally be discriminated and are regarded as having little

evidential value. Little or no published data exist on the evidential value of cotton fibers of other

colors,” and in their study of cotton fibers using both visual and spectral methods of analysis they state that, “[b]lue denim fibers are already known to be basically indistinguishable over the

visible spectral range, which makes them evidentially valueless in most instances” [20].

23 Synthetic fibers are generally discriminated based on their generic type, diameter or width,

surface morphology, cross-sectional shape, and the presence of delustering agents, their particle

size, and distribution. Unfortunately these comparative microscopy techniques are valueless

when it comes to discriminating cotton fibers. Due to the irregular shape and twists of cotton

fibers, the cross-sectional shape varies as well as the width, and natural fibers also do not contain

delustering agents [20].

Fiber dyes are often studied in the hopes of discriminating cotton fibers. In 1988, Grieve,

Dunlop, and Haddock analyzed non-denim red, black, and blue cotton fibers using

microspectrophotometry, fluorescence examination, a comparison microscope, and in some cases

thin-layer chromatography in order to assess their evidential value. The red, black, and blue

cotton fibers were chosen for analysis because the experience of the author’s found that they

were most common in casework. Forty-six samples of each color were analyzed using

comparison microscopy and microspectrophotometer and compared with one another for a total

of 1035 comparisons for each color. Ten absorption spectra replicates for each sample were

collected using the microspectrophotometer. Pairs of samples that were considered a “match” based on comparison by the microscope or their absorption spectra from the microspectrophotometer were then compared using fluorescence microscopy. The results of

these analyses showed that using comparison microscopy alone was not enough to discriminate

fully between colored cotton fibers. Approximately one in three red cotton fiber samples would

be expected to match as well as one in seven blue and one in four black cotton samples. The

result of the microspectrophotometer analysis proved to discriminate very highly and was best

for the blue fibers, followed by the red, and the black fibers had the least amount of

24 discrimination. It was also noted that the black cotton fibers appeared grey or dark grey under the microscope, and some exhibited greenish or bluish in color. When microscopic comparison was used in conjunction with the microspectrophotometer and fluorescence analysis, “the chances of finding a match from randomly chosen [non-denim] blue cottons [were] of the order of 1 in 260, 1 in 150 for red, and 1 in 60 for black” [20]. It was also noted that some of the pairs of matching samples were from the same manufacturer.

Grieve, Dunlop, and Haddock also collected thirty denim samples and compared them using comparison microscopy and the microspectrophotometer. All of the denim samples were visually indistinguishable with the exception of one which was a mixture of cotton and polyester fibers with a different type of dye that the authors speculated may be suitable for simultaneous dyeing. The blue cotton fibers in this sample matched eight other non-denim samples visually; however, the spectra did not match. Only one of the non-denim samples (from the collection of red, black, and blue cotton fibers) matched any of the denim fibers; therefore, the authors concluded that the blue denim fibers have their own unique characteristic color in comparison to other blue non-denim cotton fibers. The authors state that nearly all of the blue denim cotton spectra fall within a particular range, resulting in a very low evidential value for blue denim [20].

According to the European Fibres Group (EFG), fibers originating from denim material containing indigo dyes have so little evidentiary value that they are rarely examined. However, the EFG states that indigo substitutes have also become more popular [28].

2.2 Oligosaccharide Analyses and Differentiation of Cellulosic Materials

25 2.2.1 Work by Allen K. Murray

In 1996, Murray and Brown published their work on glycoconjugate analysis at the Beltwide

Cotton Conference. They define glycoconjugates as “carbohydrates covalently linked to other carbohydrates, proteins or lipids”. Murray and Brown analyzed the glycoconjugates from fifteen

varieties of developing cotton and were not able to find significant differences in the varieties

[8].

In 2002, Murray also analyzed various species of woods and different paper products using high

pH anion exchange chromatography with pulsed amperometric detection (HPAEC-PAD). The oligomers were extracted from the cellulosic materials using weak acid, and were determined to be reducing sugars that contain galactose, glucose and mannose. Additional analysis on the extracted oligomers concluded that the difference between successive oligomers was not a difference in mannose or galactose but additional glucose units. Murray claimed that the presence of specific oligomers and their quantitative distribution are “unique to each plant and tissue as well as to the developmental state” [9]. After prolonged wear and washings, the quantity of oligosaccharides and oligomers extracted from cotton textiles was reduced [29]. The oligosaccharides and oligomers from twenty-two different species of woods were analyzed, and it was observed that no two species of wood displayed identical chromatograms. The differences between the woods were hypothesized by Murray as being the result of differences in growth patterns and biochemistry. Teak wood was analyzed, and recently harvested teak was compared to weathered teak from a nineteen year-old sea-going vessel. The two teak chromatograms were observed to be “almost indistinguishable” with the exception that the weathered teak

26 chromatogram had a response that was less intense than the recently harvested teak. The need

for the expansion of the chromatogram was suggested to be a result of the oligomeric material

from the old teak “being extracted by repeated exposure to both salt and fresh water as well as

exposure to sunlight” [9, 29]. In addition to the oligosaccharide analysis, UV absorption at 280

nm was used to detect the presence of phenolic compounds in the woods which may be from the

phenolic amino acids in proteins or from components in lignin. For the paper products, every product tested appeared to have a unique chromatograph, which was hypothesized as resulting from different cellulose sources and processing (i.e. type and degree of bleaching, surface

finishing, and coloring) [9, 29]. An additional set of experiments was performed by extracting

the oligomers in the presence of substrates such as glycerol, m-insitol, sucrose, raffinose, and

cellobiose. The addition of these substrates reduced the amount of extracted oligomers,

suggesting that their naturally occurring content may influence the amount of extractable

oligosaccharides [29].

Murray also analyzed oligomers extracted from cotton fibers using an enzyme extraction

followed by incubation with cellulose or beta-glucosidase. Murray concluded that there were

“striking differences between cotton fibers from different varieties with respect to their susceptibility to enzymatic degradation” [30].

Glycan oligomer analysis was performed on mature cotton fibers from six locations in Brazil,

and the chromatograms of the glycan oligomers were determined to be distinct for the cotton

species, plant tissues, and the plant’s developmental stage. The glycans were extracted using

cold water, 0.1 N HCl, and 80% acetic acid in 1.8 N nitric acid [31].

27

Cotton cultivar “Ultima” (Gossypium hirsutum), grown in California was analyzed by a water extraction in which 0.5 mL of water was added to 5mg of finely diced raw cotton or was homogenized, and sonicated in ice water for 15 minutes before being centrifuged for 5 minutes.

Glucose, fructose, sucrose, ribose, galactitol, mannitol, arabitol, raffinose, stachyose, and verbascose were found in the supernatant using HPAEC-PAD analysis. After the supernatant was removed, one milliliter of 0.1 N HCl was added to the sample, mixed, and boiled in a water bath for thirty minutes. This extraction released the glucose containing oligomers. After the supernatant was removed, a third extraction was conducted by adding one milliliter of 80% acetic acid that was 1.8 N in nitric acid, mixing, and placing in a water bath (100 °C) for thirty minutes. The third extraction also released glucose containing oligomers. The chromatograms of the last two extractions differ in that more oligomer material is removed by the acetic/nitric extraction than the HCl extraction for mature fibers. Immature fibers (i.e. 21 days post-anthesis) provided more oligomer material in the HCl wash. Murray concluded that the relative concentrations of the oligomers extracted were characteristic of the source of the cotton [32].

Glycoconjugates are comprised of a monosaccharide conjugated to at least one additional monosaccharide and optionally to a protein or a lipid. Using the same analytical method previously mentioned, Murray claims the “pattern of multimers is indicative of the origin of the cellulosic material (e.g., the plant species the material comes from) as well as quality and processing state of the material”. Murray noted that as fabric undergoes wear and washings, the amount of oligomers extracted decreases. He also claimed that the multimer patterns were unique for different species of wood and also measured the UV absorbance (at 280 nm) of the

28 phenolic amino acids in the proteins and lignin found in wood. Murray noted that older teak

wood had almost no absorbing compounds in comparison to the newer teak wood, but the

chromatograms of the multimers from the old and new teak were identical if the old teak scale

was amplified 20 times. Murray speculated that the oligomers were extracted from the old teak

due to the wood’s exposure to fresh water, salt water, and sunlight. Murray also claimed that

different paper products have a unique “fingerprint” that may be a result of the degree of

processing [33].

Chromatography was performed using HPAEC-PAD using a CarboPac PA-1 column.

“The eluent was 150 mM sodium hydroxide, isocratic from 0 to 5 [minutes] then a linear sodium acetate gradient from 5 to 40 [minutes] going from 0 to 500 mM in 150 mM NaOH at a flow rate of 1 ml/min. The detector wave form was the following: 5-0.50 s, 0.1 V; 0.51-0.59 s, 0.6 V; 0.60-0.65 s, -0.6 V; integration 0.30-0.50 s. For monosaccharide composition, oligomers were obtained by collecting fractions of the HPAEC-PAD eluent, which was passed through a Dionex ASRS-II anion suppressor to remove salt. Fractions were then lyophilized and taken up into 200 μL of water, made up to 2N trifluoroacetic acid, flushed with argon and sealed in screw cap plastic vials with O-rings. The samples were then placed in a heating block at 100°C for 2-4 [hours]. Following hydrolysis, the samples were taken to dryness in a Speed-Vac overnight and then taken up to 200 μL of water for HPAEC-PAD on a Dionex CarboPack-PA10 column under isocratic conditions in 15 mM NaOH” [29].

Analysis was conducted on Gossypium hirsutum var. DP-50 grown in the Mississippi Delta and

Sacramento Valley. The cotton bolls were freeze dried as soon as possible after collection. First a water extraction was performed, at 0°C using 0.5 mL of 18.3 megaohm water, 20 mg of sample, and sonicating, in order to remove the soluble monosaccharides and oligosaccharides.

After removing the supernatant, an acid extraction with 0.5 mL of 0.1 N HCl was performed at

100°C for 30 minutes that extracted the glucose containing oligomers. The extracted oligomers

29 were neutralized with 1 N NaOH, and then analyzed with HPAEC-PAD. When extracts were

performed on fiber, wood, and paper products, 40-60 mg of sample was used along with 1.0 mL

of water and HCl [29].

Murray also extracted the oligomer-protein colloid following the cold water wash by adding 0.5

mL of water and 50 μL of toluene (used to prevent microbial growth), incubating in a 37°C water

bath for 24 hours, then centrifuging and filtering the supernatant. The oligomers were released using protease treatment or weak acid extraction. The proteases tested for oligomer extraction were proteinase K, trypsin, chymotrypsin, and Pronase®; however, chymotrypsin was the most effective [29].

Following the extraction of the multimers, enzyme treatments with cellulase (Trichoderma

reesei) or β-glucosidase was performed using 1 mg/mL of enzyme in 50 mM sodium acetate

buffer, at pH 4.8. The latter increased the abundance of multimer peaks and produced an

additional peak at a retention time just longer than 20 minutes. This was speculated as being a

result of the trimming of higher order oligomers that were not resolved on the column or that had

a low detector response. The cellulase eliminated most of peaks with longer retention times and

increased the first peaks in the multimer series. The cellulase treatment released only glucose

monosaccharides [29].

2.2.2 Additional Methods of Analyses

30 Previous chromatographic methods that have been used to analyzed saccharides and

oligosaccharides include ligand-exchange, size exclusion, and hydrophilic chromatography using

a silica-based polar stationary phase. Underivatized oligosaccharide analysis requires specific

stationary phases such as alkylated silica-bonded, amino-bonded, or various ion-exchange

columns [34]. Liu et al. analyzed underivatized oligosaccharides up to 11 glucose units in length using high-performance liquid chromatography (HPLC) coupled with electrospray ionization mass spectrometry (ESI-MS) [34]. The oligosaccharides were detected as low as 50 pg through the use of a Cyclobond 12000 beta-cyclodextrin stationary phase column with a mobile phase containing a high concentration of acetonitrile. This method has high chromatographic resolution, low mass spectrometric detection limits sensitivity, and high column stability. The acetonitrile-rich mobile phase allows hydrogen bonding to occur between the stationary phase cyclodextrin hydroxyl groups and the hydroxyl groups of the oligosaccharides.

Liu et al. demonstrated that the addition of a small amount of an acidic additive (formic acid or

acetic acid) enhanced the MS sensitivity in the positive ion mode, along with a smaller internal

column diameter (1 mm I.D. versus 4.6 mm), and the mobile phase flowing at a rate of 200

μL/min versus a higher flow rate (up to 800 μL/min). The lower flow rate was speculated to

reduce the size of the charged droplets during the electrospray process; therefore, the ions were

more efficiently released into the gas phase [34].

Cellobiose and maltose both contain two glucose units; however, cellobiose is connected by a

beta-(1,4)-glycosidic linkage, while maltose has an alpha-(1,4)-glyclosidic linkage. The two

could not be differentiated by Liu et al. based on their chromatographic retention times using a

31 cyclodextrin column, but sucrose did have a shorter retention time and lactose a slightly longer

time [34].

Electrospray ionization coupled with mass spectrometry (ESI-MS) was used to identify and

characterize maltooligosaccharides in beer samples [35]. Carbohydrates are non-volatile

compounds; therefore, they are easily analyzed using high-performance liquid chromatography

(HPLC) [35]. Refractive index detection (RID) and pulsed amperometric detection (PAD) have

been used for carbohydrate analyses; however, RID lacks selectivity and sensitivity and PAD

requires a time consuming cleaning procedure to remove any interferences from the sample

matrix prior to analysis. Capillary electrophoresis has also been used, but the carbohydrates

must be derivatized first or “highly alkaline electrolytes can be used to ionize underivatized

carbohydrates making them suitable for analysis by capillary electrophoresis with indirect UV

detection” [35]. Other methods used to study the structures of carbohydrates involve using ESI-

MS and matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass

spectrometry.

As Won et al.. point out in their paper, sugar compounds are not easily protonated to form

[M+H]+ ions in the positive ESI mode; therefore, adduct complexes are formed between the

analytes and inorganic cations, such as ammonium, sodium, or lead cations [36]. Won et al. determined typically more than one lead (Pb2+) ion would attach to the sugar analyte, resulting in

+ + decreased detection sensitivity; however, the sodium (Na ) and ammonium (NH4 ) cations were

readily detected. Won et al. tested the efficiency of adduct formation using both 5 mM sodium

chloride (NaCl) and 5 mM ammonium acetate (NH4Ac), and concluded that the ammonium

32 acetate solution formed adducts more efficiently since it was more volatile than the sodium chloride solution and required less ESI source cleaning [36].

In a paper by Choi and Ha, the author’s characterized maltooligosaccharides using sodium cation

adducts and observed that besides the [M+Na]+ ion detected, the [M-OH]+ ion was the second

most abundant ion [37]. Mauri et al. analyzed carbohydrates in beer samples using flow

injection into ESI-MS without chromatographic separation or derivatization to obtain a

fingerprint of the different maltooligosaccharides present in beer samples. The ions detected for

maltose, maltotriose, maltotetraose, maltopentaose, maltohexaose, and maltoheptaose were

[M+Na]+ and [M+K]+, where the sodiated adducts were the most prevalent, and the potassium

source came from the beer. Matrix effects were analyzed by adding 0-20 mM of sodium,

potassium, and cesium ions to a solution of maltotriose and maltotetraose standards (150 μg/mL) and to diluted beer samples. The maximum MS response occurred after adding 3 mM of sodium ions and approximately 1 mM for potassium and cesium ions. Cation concentrations greater than

5 mM decreased the ion abundances. The authors reported that different oligosaccharide patterns

were observed for each sample spectra for two different 100-fold diluted beers [35].

The characteristic maltooligosaccharides fingerprints of beer were also investigated using direct

flow injection with ESI-MS and ESI-MS/MS by Araujo and his group [38]. Twenty-nine beer

samples (classified as either ales or lagers with sub-classifications: pilsener, pale ales, bock,

stout, porter, mild ale, and malt) from Brazil, Europe, and the USA were analyzed in both

positive and negative modes using a Micromass quadrupole time of flight (Q-TOF) mass

spectrometer. The ESI-MS conditions were as follows: 100°C source temperature, 120°C

33 desolvation temperature, 3.0 kV capillary voltage and 40 V cone voltage. Formic acid was

added to the samples analyzed in the positive mode, and ammonium hydroxide was added for

negative ion analysis. Sodium and potassium adducts were observed in the positive mode, and

[M-H]- and chloride adducts were analyzed in the negative mode. Principle component analysis

(PCA) was performed on the results and the samples were grouped into three major types: pale colored (pilsener, lager, and pale ales), dark colored (bock, stout, and mild ales), and malt

(“Malzbier”) beers [38].

Graphitized carbon solid phase extraction (SPE) was used to separate sugars in a dilute solution

from salts, alkali, or mineral acid which may be present in the solution [39, 40]. Glucose was not

retained by the activated carbon, but rather eluted with water; however, glucose oligosaccharides

were retained by the activated carbon and eluted using water with the addition of an organic

modifier, such as acetonitrile [39]. Alltech Associates Carbograph (non-porous graphitized

carbon) cartridges with a bed weight of 150 mg and a volume of 0.5 mL were used and

conditioned by washing with three 1.5 mL aliquots of acetonitrile or methanol, followed by three

1.5 mL aliquots of water [39]. Batch elutions of the oligosaccharides were performed with water

containing 1-butanol as an organic modifier. Another eluent used was 1:1 acetonitrile-water to

remove raffinose from a column. Dextran 500 was heated (100°C) in 0.1 M hydrochloric acid

for two hours, then cooled and diluted with water before being applied to a conditioned

graphitized carbon cartridge. After rinsing with water, the oligosaccharides were eluted with sequential batch elutions butanol-saturated water (in varying concentrations). The elution of N-

acetylneuraminyllactose was also performed with 1:3 acetonitrile-water [39]. Packer et al. used the Carbograph graphitized carbon SPE column as well, and conditioned it by washing with

34 three column volumes of 80% (v/v) acetonitrile in 0.1% (v/v) TFA followed by three column

volumes of water. After passing an aqueous solution containing oligosaccharides and salts

through the column (at a rate of 0.5-1 ml/min), the salts were washed off the column using three

column volumes of water. The oligosaccharides were then eluted batch wise with three column volumes of acetonitrile and water or dilute acid. Neutral oligosaccharides were released from the

Carbograph SPE column in another experiment using 25% (v/v) acetonitrile in water, and acidic

oligosaccharides (oligosaccharides which are sialylated, sulfated, or phosphorylated) were eluted with 25% (v/v) acetonitrile in 0.05% TFA [40].

Oligosaccharides are also detected using methods such as fluorescence by first derivatizing the

oligosaccharides in order to attach a chromophore or fluorophore. Derivatized that have been

used for monosaccharides and oligosaccharides include 1-aminopyridine (AP), 4-aminobenzoic

acid ethyl ester (ABEE), 1-phenyl-3-methyl-5-pyrazolone (PMP), and PMP’s methoxy analogue

(PMPMP) [41]. Successful analysis of oligosaccharides using ESI was also achieved by

derivatizing the oligosaccharides with a brominated aromatic amine reagent, 2-amino-5-

bromophyridine (ABP) [42].

It has been stated that oligosaccharide analysis using ESI-MS produces a relatively weak signal for native oligosaccharides, due to the hydrophilicity of oligosaccharides limiting the surface activity in ESI droplets; however, the sensitivity may be significantly increased using nano-ESI

[43]. The detection may be increased further by derivatizing the oligosaccharides which reduces their hydrophilicity and increases their partitioning to the droplet surface and thus gaseous ion

35 formation [43]. Matrix-assisted laser desorption/ionization (MALDI) has also been used for oligosaccharide analysis [43].

ESI-MS conditions may be manipulated in order to favor the production of desired ions,

including protonated or metal adduct ions like ammonium or sodium adducts [43]. “Ammonium

adducts formed in the presence of ammonium acetate readily decay to protonated ions due to the

energy required to desolvate ions during the electrospray process” [43].

Another study conducted by Gilbert and Kakot concluded that cellulosic fabrics may be

discriminated by diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy and

applied chemometrics. This study demonstrated that discrimination was possible based on fabric

dye, fabric type, and textile processing. Since the cotton-cellulose polymer consists of linked

cellobiose units, the IR spectra was used to reflect the structural features and modifications

resulting from differences in the fiber and fabric processing and structure. Using DRIFT spectra,

all spectra of the cellulosic material (bleached muslin, bleached voile, poplin, linen, modal,

viscose, and blue, green, or navy poplin) were similar in the 1800-800 wave number range. The

dyed poplin had peaks between 1450 and 1700 wave numbers, and the modal and viscose fabrics

had a broad shoulder between 1400-1430 wave numbers and a change in two peak intensities at

1170-1200 wave numbers. While the rest of the spectra appeared to be similar, principle

components analysis (PCA) analysis using PC1 vs. PC2 provided nine clusters that corresponded

to the nine fabric types. PC1 differentiated most poplin types, and PC2 discriminated between

the green and navy poplin spectra. The authors mention that their previous work was capable of

differentiating between undyed cotton-cellulose fabrics (i.e. cotton-poplin, muslin, and voile)

36 using DRIFT sampling in combination with PCA, soft independent modeling of class analysis

(SIMCA), and fuzzy clustering (FC) [44].

Elemental analysis isotope ratio mass spectrometry (EA-IRMS) was utilized by van der Peijl to

study 20 different brands of blue jeans (denims). The isotope ratios for hydrogen (δ2H), carbon

(δ13C), and oxygen (δ18O) were measured in triplicate for each sample. The results demonstrated that most of the samples were capable of being discriminated based on the wide ranges and despite large standard deviations; therefore, the van der Peijl concluded that laser ablation IRMS

(LA-IRMS) may be an adequate technique for comparing single fibers [45].

According to Martsinkovskaya et al., differentiation of varieties and species of Gossypium cotton are possible by ribosomal intergenic sequencing [46]. Whereas, Suzuki et al. analyzed single

fibers dyed by indigo and its seven derivatives: 4,5,4’,5’-tetrachloroindigo, 5-bromoindigo, 5,5’-

dibromoindigo, 5,7,5’,7’-tetrabromoindigo, 4,5,7,4’,5’-pentabromoindigo, 4,5,6,4’,5’,6’-

hexabromoindigo, and 7,7’-dimethylindigo (Refer to Figure 5).

The dyes were analyzed using microspectrophotometric examination of transmittance spectra in

the ultraviolet-visible (UV-Vis) region (240 nm-760 nm) [47]. The samples were embedded in

glycerin and a small amount of intra-sample variation in the transmittance intensity of woolen

fibers was observed, but there was a relatively large inter-sample variation among the fibers from

different areas in the textile. The study claims that sample fibers dyed with indigo and its seven

derivatives could be distinguished from one another when comparing their ultraviolet-visible

transmittance spectra lambda max, lambda min, and minute shoulder peaks [47].

37

2.3 Instrumentation

2.3.1 LC-ESI-MS

2.3.1.1 Liquid Chromatography

In chromatography, the flow of the mobile phase causes the analyte to pass through the

stationary phase, where the mixture of analytes is separated based on their affinity for, and

partitioning between, the stationary and mobile phases. In liquid chromatography (LC), a porous

solid stationary phase is used along with a liquid mobile phase. The time it takes for an analyte

to elute through the column, known as the retention time, is dependent on the flow rate of the

mobile phase and the dynamic equilibrium of the analyte between the stationary and mobile

phases. LC columns are typically composed of metal and are generally operated at ambient

temperatures. If a single mobile phase is used, the chromatography is isocratic; whereas, if

multiple mobile phases are used to create a mixture of solvents where the concentration

gradually changes, the process is known as gradient elution. As the analytes elute from the

chromatography column, they are sensed by a detector which sends an electrical signal to a

readout device or recorder and produces a chromatograph. The chromatograph shows the

detector response (y-axis, with the peak areas being proportional to the amount of sample

detected) that emerges from the column versus the retention time of the analyte (x-axis). Liquid

chromatography is an excellent technique for the analysis of nonvolatile and/or non thermally stable analytes such as sugars, peptides, proteins, and nucleosides [48].

38

In order to couple liquid chromatography (LC) with mass spectrometry (MS) an atmospheric pressure ionization (API) interface is used (Refer to Figure 6).

Figure 6: LC-API-MS Schematic

The API-MS techniques are useful because they can handle the liquid solvents used in LC, allow for analysis of nonvolatile, polar, and thermally unstable compounds that are typically analyzed using LC, the sensitivity is comparable to the GC-MS, and the API systems are simple to use and very rugged [49]. There are three main API techniques: electrospray (ES), pneumatically assisted ES, and atmospheric-pressure chemical ionization (APCI). The ES technique uses electrical fields to generate charged droplets and analyte ions using a desorption ionization process. Pneumatically assisted ES is the same as ES, but the initial droplet is formed using pneumatic nebulization, and the APCI technique involves “a gas-phase ionization process initiated by a discharge sustained by the LC mobile phase vapor” [49]. The sensitivity of liquid chromatography coupled with atmospheric pressure ionization and mass spectrometry (LC-API-

MS) may be increased by solvent choice, flow rate, solvent additives, and pH which influence

LC separation and enhanced ionization. Post-column techniques may also be used to enhance ionization without changing the chromatographic conditions [49]. When coupling the LC to a

MS detector, common LC additives such as sulfate, phosphate, and borate buffers must be replaced with mobile-phase additives that are volatile. The use of volatile solvent additives, like

39 those in Table 2, prevents contamination of the API chamber or plugging of the sample orifice.

ESI also requires that the solvent additives do not form strong ion pairs; otherwise, the ions may

be neutralized after desorption [49].

Table 2: Solvents and Additives Suitable for API-MS Application Solvents and Additives pH Acetic acid, formic acid, trifluoroacetic acid for ES positive- ion detection; ammonium hydroxide for ES negative-ion detection (typically in the 0.1-1% range) Buffers, Ion-pair Reagents Ammonium acetate, ammonium formate, triethylamine heptafluorobutyric acid, tetraethyl or tetrabutylammonium hydroxide (10-11 mM level) Cationization Reagents Potassium or sodium acetate (20-50 μM level)

Solvents for API-MS Methanol, ethanol, propanol**, isopropanol**, butanol**, acetonitrile, water, acetic acid*, formic acid*, acetone*, dimethylforamide*, dimethyl sulfoxide*, 2-methoxy ethanol**, tetrahydrofuran*, chloroform* Solvents for APCI Only (Not Hydrocarbon solvents (i.e. hexane, cyclohexane, toluene), CS2, Suitable for ES) CCl4 * Solvents used in the 5-20% range. ** Best for negative-ion operation [49].

There are various chromatographic techniques that are capable of being coupled to API-MS such

as reverse phase, normal phase, ion-pairing, size-exclusion, ion exchange, or immunoaffinity

separations. Reverse phase chromatography was used for the experimental procedure discussed in this thesis. Reverse phase chromatography separates analytes based on the partitioning of the analyte between the liquid mobile phase and the stationary phase. The pH of the mobile phase is usually controlled to aid in the separation of neutral compounds.

2.3.1.2 Electrospray Ionization

40 There are several ionization methods that are capable of producing gas phase ions of fragile

compounds. These “soft ionization” methods include field desorption (FD), fast atom

bombardment (FAB), secondary ion mass spectrometry (SIMS), electrospray ionization (ESI),

and matrix-assisted laser desorption/ionization (MALDI) [50]. “These methods are called ‘soft’

because, when proper experimental conditions are used, intact molecular ions can be produced

with minimal fragmentation” [50]. Fragmentation may be induced in electrospray ionization by

“increasing the acceleration voltages in a medium pressure region after droplet formation” [50].

Carbohydrates have been shown to be analyzed well using ESI-MS [49].

The electrospray ionization (ESI) interface allows the liquid samples to be transferred to the gas

phase in order to be analyzed in the mass spectrometer. The ESI technique is a very useful

method of ionization because it allows samples to be introduced in solution, and it is a soft

ionization technique which allows for the analysis of the structure of easily ionized molecules and molecule-ion complexes and the non-covalent interactions between molecules. The ESI technique provides the ability to analyze molecules with large molecular weights by forming multiply-charged ions [48, 49]. Electrospray mass spectrometry was introduced in 1984 by

Yamashita and Fenn, and around the same time Aleksandrov et al. independently came up with a

similar approach. The idea of using electrospray as a source of gaseous ions for mass spectral

analysis was first proposed by Dole in 1968, but his experiments were limited in scope [49].

In the electrospray process, the sample solution is passed through a capillary that has an electric

field applied to the tip. As the solution is eluted from the capillary, the electric field causes the

formation of a dipolar layer at the liquid’s meniscus. The electric field applied to the tip of the

41 capillary is dependent on the solvent used; therefore, a greater electric field must be used for

solvents with a higher surface tension (i.e. water) [49].

When turned on, the [electric field] will penetrate the solution at the capillary tip and the positive and negative electrolyte ions in the solution will move under the influence of the field until a charge distribution results which counteracts the imposed field and leads to essentially field-free conditions inside the solution [49].

When the capillary is positive, the negative electrolyte ions in the solution are attracted to the capillary leaving the positive electrolyte ions to form the double layer near the surface. A Taylor

Cone is formed with the help of a nitrogen drying gas, and positively charged droplets are

released from the cone. The flow rate of the solvent effects the size of the Taylor Cone and corresponding droplets, and solvent evaporation leads to shrinkage of the charged droplet size

[49]. Refer to Figure 7 for an illustration of the process.

42

Figure 7: ESI Spray Schematic

There are two main theories proposed that describe how the gaseous ions are formed from the

charged droplets, and both theories explain observed mass spectral features. The “Charged

Residue Model” (CRM) was originally proposed by Dole and Röllgen in 1968 and 1970. The

CRM theory was based on a theoretical paper published by Lord Rayleigh in 1882 in which Lord

Rayleigh presented a theory on “what would happen as solvent evaporates from a droplet of

volatile liquid containing an excess of either anions or cations” [51]. He proposed that the ions would be located at equidistant intervals along the surface of the droplet due to the repulsive forces between the charges. As the solvent from the droplet evaporates, the surface ions would be forced closer together until they reached what is now referred to as the “Rayleigh Limit”. At

43 the Rayleigh Limit, the coulomb repulsion forces between the surface ions exceed the surface

tension of the droplet liquid and the droplet breaks into smaller offspring droplets resulting in an increased available surface area. This process could continue until “ultimate droplets” are formed where each contained only a single solute molecule, then the last of the solvent would evaporate to leave a gas-phase solute ion [51]. The second theory of gaseous ion formation was suggested in 1975 by Iribarne and Thomson, and it is called the “Ion Evaporation Model” (IEM).

Iribarne and Thomson proposed that, rather than the charged droplets becoming small enough to contain a single charged analyte, the charge density on the droplet surface would become so great in droplets with a radius of less than 10 nm that one or more of the surface ions would be

“pushed” into the ambient gas [49, 51]. As the solvent continued to evaporate on the original charged droplet, this process would continue until most of the charged surface ions were expelled into the ambient gas [51].

In January 2007, the results of further investigation into ESI ion formation by Steve Nguyen and

John B. Fenn were published in the Proceedings of the National Academy of Sciences. The

authors concluded:

“[F]or most, if not all, cases in which ESI is effective, gas-phase solute ions are formed from charged droplets according to the sequence of events described in the IEM of Iribarne and Thomson. However, in the case of very large parent species, including [polyethylene glycols] with molecular masses as high as 5,000,000 Da, [it is believed] that the CRM of Dole et al. may comprise the more likely ionization scenario” [51].

44 Nguyen and Fenn also state that work conducted by Kebarle et al. concluded that metal ions (i.e.

Na+, K+, or Cs+) are likely produced by IEM, and that denatured proteins produce larger ions

through CRM [51].

It is possible to form adducts between neutral analytes and anions or cations in the solution in

order to create a charge molecule that may be detected using a mass spectrometer [49]. The

ionization efficiencies vary for different compounds; therefore, the compounds will have

different mass spectral responses. If a mixture of analytes is to be ionized in ESI, some compounds may ionize more efficiently, which may result in the failure to detect some analytes

[52]; therefore, separation of the analytes using liquid chromatography is ideal. The pH of the solution may also affect the detection of the analyte by aiding in the protonation or deprotonation of the analyte. For example, when positive ions are desired, a solution with a lower pH may aid in the protonation of the analyte [49].

The amount of fragmentation of the analyte ions depends on the internal energy received from

the source [50]. Due to the ability to produce multiply charged ions, charge distributions may be

shifted, fragmentation modified, and weak bonds broken resulting in ESI-MS spectra that are

difficult to reproduce and compare [50].

While there are other detectors that may be used with LC, such as ultraviolet absorption detectors

or refractive index detectors, the mass spectrometer has been established as a very powerful tool.

Coupling the electrospray to the quadrupole ion trap (QIT) initially posed a challenge, because

ESI occurs in atmospheric pressure (approximately 760 Torr) while the QIT operates under a

45 vacuum (See section 2.3.1.3.1 Quadrupole Ion Trap for details about the QIT). The ES-QIT

mass spectrometer (LCQ™) designed by Finnigan is shown in Figure 8.

Figure 8: Finnigan Corporation LCQ ESI-QIT-MS

-3 -5 P2, P3, and P4 are operated at pressures of 1 Torr, 1 x 10 Torr, and 2.5 x 10 (uncorrected) Torr, respectively, and are maintained by a vacuum system composed of a rough pump and a dual- port, turbomolecular pump. In the P1 region, the analyte in solution is electrosprayed from a

stainless-steel needle at approximately 4.5 kV for the positive ion mode and is sampled by the

heated capillary. Generally the heated capillary is operated at 200°C (150°C was used for the

experimental process in this thesis) and is 11.5 cm long with an internal diameter of 400 μm.

The molecules are expelled from the heated capillary as a supersonic free jet of solvent and

analyte molecules in the P2 region. The molecules are then passed through a skimmer using a

tube lens as a gating element. After the ions pass through the skimmer, they are collected by the

46 first rf octopole, located in the P3 region, then passed to the second rf octopole through an interoctopole lens located in region P4. Both the rf octopoles are 5 cm in length and are operated at 2.5 MHz and 400 Vpp. Finnigan’s LCQ™ design also positions the second octopole inside the entrance end cap of the QIT. The QIT is operated at a pressure between 1x 10-3 and 2 x 10-3 Torr

[49].

2.3.1.3 Mass Spectrometer

A mass spectrometer (MS) performs three basic functions. After the sample is injected into the

MS, the sample is ionized, the ions are separated according to their mass-to-charge (m/z) value in the mass analyzer (the ion trap), and finally the ions are detected (Refer to Figure 9).

Figure 9: MS Schematic

A spectrum is produce which plots the abundance of the ions (y-axis) versus the m/z value of the ion (x-axis); therefore, a single chromatogram may contain hundreds of mass spectra.

2.3.1.3.1 Quadrupole Ion Trap

47 The quadrupole ion trap (QIT) mass analyzer was first publicly described in 1953 by Wolfgang

Paul and H. Steinwedel at the University of Bonn. The QIT was used as a scanning mass

spectrometer following the invention of the mass selective instability scan by Stafford, Kelley,

and Stephens. The ion trap mass spectrometer was first introduced commercially in 1983 by

Finnigan Corporation [49].

The QIT is composed of three hyperbolic electrodes. The two end cap electrodes are identical to

one another, and the ring electrode is donut shaped and sandwiched between the end cap

electrodes. The distance between the end cap electrodes and the ring electrode is maintained by

spacers made of ceramic or quartz.

Each end cap electrode has a hole in the center which allows the passage of ions into and out of the ion trap; therefore, one end cap electrode has a hole that allows ions to be injected into the trap (i.e. from the electrospray source), and the other electrode has a hole that allows for the ejection of ions from the QIT to the detector. In order to inject only positively charged ions, a small negative dc offset is applied to each of the electrodes [49] (Refer to Figure 10).

48

Figure 10: QIT Schematic

Mass analysis using an ion trap is based on ion motion in a trapping potential well which is created when a radiofrequency (rf) potential is applied to the ring electrode and the two end-cap electrodes are grounded (with the exception of oscillating potentials of low amplitude) [53]. An oscillating rf electric field directs the ions away from and back to the center of the trap in both the axial (z, distance between the end-cap electrodes) and radial (r, diameter of the ring electrode) directions. When a positive rf voltage is applied, the quadrupole potential well has a saddle-shaped surface which directs the ions downhill in the z-direction and wards the center of the trap. This movement is reversed when a negative rf voltage is applied. The field strength

varies linearly with the distance from the center of the trap, so the ions further from the center of

the trap are subjected to a greater force, returning the ions to the center. Therefore, the applied rf

49 voltage creates a potential well that keeps the ions trapped by alternating stabilizing and destabilizing forces [54].

In conjunction with the trapping potential well, and before resonant excitation, “ions are focused

collisionally to the vicinity of the [center] of the ion trap under the influence of collisions with

helium buffer gas atoms” at a pressure of approximately 1 mTorr, which aids in the removal of

kinetic energy from the ions [53, 54].

The ion trap is typically defined as having the geometry corresponding to the following equation

[49]:

2 2 0 = 2zr 0 (2)

However, in the early 1980s, researchers at Finnigan Corporation discovered that ion traps built

to these specifications had a poor mass accuracy; therefore, the end caps were moved outward by

a factor of 0.11 z0 which “improved the homogeneity of the quadrupole field near the center of

the trap where the ions are stored” [49]. For the stretched ion trap in the Finnigan LCQ and

GCQ, r0=0.707 cm and z0=0.785 cm [53]. The Finnigan LCQ™ is also capable of achieving “a

mass-to-charge ratio of 2000 Dalton/charge by using a V0-p of 8.5 kV, a r0 of 0.707 cm, z0 of

0.783 cm, a frf of 760 kHz, with a qz-eject value (see below) reduced to 0.83” [49, 53, 54].

The trajectories of the ions in the trap are predicted based on second-order Mathieu differential

equations; however, the trajectories are also influenced by the effects of space-charge repulsion,

50 the imperfect field, and collisions with gasses such as helium [49]. Equations 3 and 4 represent

the ion trajectories, in terms of the Mathieu parameters az and qz [53].

−16eU a = (3) z 2 2 2 (0 2zrm 0 )Ω+

8eV q = (4) z 2 2 2 (0 2zrm 0 )Ω+

where the radial direction is represented by r, the axial direction by z, the dc amplitude by U, the rf amplitude by V, the ion charge by e, the ion mass by m, the inner radius of the ring electrode by r0, the axial distance from the center of the trap to the nearest point on one of the end cap electrodes by z0, and Ω = 2πfrf where frf is the frequency of the main rf voltage [49]. In most

commercial ion traps, the dc potential is not applied to the electrodes; therefore, az = 0 and the

ion trap is operated based on the qz axis [53].

The Mathieu parameters, a and q, denote the stability and motion of the ion within the quadrupole field and may be summarized by the QIT stability diagram (Refer to Figure 11).

51

Figure 11: QIT Stability Diagram

The QIT stability diagram is formed when the Mathieu parameters are plotted and overlapped with the r and z dimensions of the ion trap. The operating line, az = 0, is representative of the points (az, qz) in the stability diagram that the ions must pass through until they are ejected from the trap. The operating line begins at (az, qz) = (0, 0) and passes through (az, qz) = (0, 0.908).

Near the edge of the stability diagram, at qz-edge = 0.908, the ion becomes unstable and is ejected from the ion trap through the hole in the end cap. Higher mass-to-charge ratio ions have a lower qz value than low mass-to-charge ratio ions, because qz is inversely proportional to mass-to- charge (Refer to Equation 4, shown above). As the rf amplitude (V) applied to the ring electrode is increased, the ions positioned along the operating line are moved to a higher qz value, until they reach the qz-edge value and are ejected. With the lowest mass-to-charge ratio ion being the first to be ejected from the ion trap [48, 49, 53, 54].

52 In order to measure the mass-to-charge ratio of the trapped ions, the ions must be ejected from the trap. This is accomplished by “tipping” the potential trapping well to release ions in order of ascending mass-to-charge ratio. The well may be “tipped” by linearly ramping the amplitude of the radiofrequency (rf) potential applied to the ring electrode. This process is referred to as

“mass-selective axial instability mode”, developed by Stafford et al. [53].

By applying a supplementary resonant excitation-ejection rf voltage in a dipole fashion to the

end-cap electrode, resonant excitation is performed. Each ion has a unique secular frequency in both the r- and z-directions. If the supplementary resonant rf frequency equal to the secular frequency is applied between the end-cap electrodes, the ions will gain kinetic energy and become activated in the z-direction, resulting in the ions experiencing a greater trapping field.

The ions may be ejected from the trap, if the resonant signal is strong enough. This process may be used for selective ion monitoring, selective ejection of ions, to fragment ions, or as “axial modulation”. Axial modulation works concurrently with the “tipping” of the potential well to axially eject ions from the center of the trap just before the ions reach the ejection qz value, in order to remove the lower mass-to-charge ratio ions from the trap and from the “influence of space charge perturbations induced by other ion species, so that the ions are ejected free of space charge and with enhanced mass resolution”, while leaving larger mass-to-charge ions in the trap

[54]. After the ion passes through the exit end cap electrode, it is sent to a dynode-electron multiplier detector and the results sent to the data system [49, 54].

2.3.1.3.2 Detector Response

53 The linear response of the detector to the concentration of the analyte must be tested in order to use the peak height or peak area detector responses for comparing two samples [55]. This is

generally accomplished by the construction of calibration curves which are created from

standards of known concentration versus their detector response. The response is proportional to

the amount of sample detected and is found by using the integrated peak area of the response

peak. In order to determine the linearity of a plot of concentration versus detector response, the

correlation coefficient may be calculated. A correlation coefficient (R2) equal to the absolute

value of one would correspond to a perfectly linear correlation. When the concentration exceeds a particular value, the detector becomes saturated and the detector response is no longer linear.

For this reason, the linear dynamic range (LDR) is determined for each analyte, and this region may be used for quantitation of samples with an unknown concentration [49]. For this study, integrated peak areas that fell within two standard deviations greater than the upper end of the

LDR (upper concentration) and two standard deviations less than the lower end of the LDR

(lower concentration) were said to be within the LDR.

2.3.2 GC-MS

2.3.2.1 Gas Chromatography

Gas chromatography is a technique that allows for chromatographic separation of volatile

analytes; therefore, the chromatographic column is contained inside of an oven that is capable of

being held at a constant temperature (isothermal) or a range of temperatures (temperature

programming). The gas chromatography column is generally composed of quartz and is often

54 10-50 m in length with a 0.1-1.0 mm internal diameter; hence, the column is typically called a

capillary column. The separation is based on the volatility of the analyte, where the more

volatile analytes will spend more time in the gas phase, thus being passed through the column

before the less volatile analytes, and on the affinity of the analytes for the non-solid (liquid, gum, or elastomer) stationary phase located on the interior wall of the column. The injector of a GC is typically 50°C higher in than the column oven temperature. As the analytes are forced through the column by the inert gaseous mobile phase, generally nitrogen or helium flowing between 1-2 mL/min, the analyte affinity for the stationary phase will also aid in the chromatographic separation of the analytes. For example, if two analytes have the same volatilities but their polarities differ, their varying polarities will separate them when using a polar column [48]. The analytes that have a greater affinity for the stationary phase will be retained longer by the column, thus eluting at a later time (having a greater retention time).

The eluted analytes are then sensed by the detector which produces a gas chromatograph

depicting the amount of analyte (y-axis, with the areas of the peak being proportional to the

amount of sample detected) that emerges from the column versus the retention time of the

analyte (x-axis) [48]. Refer to Figure 12 for a diagram of a basic GC instrument connected to a

detector.

55 Figure 12: GC Schematic

The sample in the gas mobile phase emerges from the chromatography column and into the ion

source of the mass spectrometer at or near atmospheric pressure with a flow rate between 0.5-3.0

mL/min. The ion source is typically operated at 10-5 mbar and 10-3 mbar for electron ionization

and chemical ionization, respectively. The change in pressure between the column and the inside of the ion source causes the gas to expand and the flow to increase to several liters per minute, so a pump is used to remove the excess gas and maintain the vacuum inside the source [48].

2.3.2.2 Mass Spectrometer

As the analytes enter the ion source from the GC, they are ionized either by an electron beam

(electron ionization) or by a reagent gas (chemical ionization), and those ions are analyzed by the

mass analyzer to produce a mass spectrum [48]. Electron ionization was used for this research

with an electron beam at 70 eV (electronvolts), and a Finnigan Ion Trap was used as the mass

analyzer.

The electron ionization process involves an electron beam that is formed when electrons are

accelerated through a 70 V electron field and gain kinetic energy of 70 eV. These electrons eject

56 an electron from the neutral analyte molecule by passing through the molecule or close to it to generate a radical-cation. Equation 5 demonstrates this process [48].

e- + M Æ M•+ + 2e- (5)

This high kinetic energy in a vacuum cause some of the radical-cations to fragment and produce a mass spectrum of ions with smaller masses, which are characteristic for each molecule [48].

Refer to section 2.3.1.3 Mass Spectrometer for details about the ion trap.

2.4 Statistical Methods of Analyses

2.4.1 One-way Analysis of Variance

A one-way analysis of variance (ANOVA), with between-subjects design is used when there is a single independent variable (Univariate) measured on a nominal scale, which may assume two or more values (categories), and a single dependent variable measured on a ratio or interval scale

[56, 57]. This analysis is similar to a student t-test; however, it is capable of comparing more than two samples. The null hypothesis of an ANOVA test states that the populations of the independent variable categories are equal on the dependent variable, or H0=µ1=µ2=…=µn. The alternative hypothesis states that there the populations of the independent variable categories are not equal on the dependent variable, or Ha=µ1≠µ2 ≠…≠µn [57].

57 Rather than comparing the actual means, the ANOVA test of significance compares the variance

between the independent variable categories to the variance within the dependent variable. The

variances are measured based on the sum of squares (SS) values. The sum of squares between

the groups (SSB), or different denim fabrics, is found using equation 6 [57].

2 = ∑ k (k − XXNSSB ) (6)

where Nk is the number of independent variable categories, X k is the mean of each of the

independent variable categories, and X is the overall mean from all of the independent variable

categories. The sum of squares within the groups (SSW), or the replicate analyses of the

individual denim fabrics, is found using equation 7 [57].

2 SSW ∑ (−= XX ki ) (7)

where X i is the individual observation. The total sum of squares (SST) may be found by summing the values for SSB and SSW.

If the null hypothesis is correct, then the variances between and within the independent variable

categories (groups) should be approximately equal; however, as the difference between the two

variances increases, the alternative hypothesis is more likely to be accepted. The comparison of

the variances is conducted based on the mean square estimates, which divides the sum of squares

58 by the degrees of freedom associated with each sum of squares. The mean square estimate between the independent variables is calculated using Equation 8 [57].

SSB MS = (8) between k −1

where k is equal to the number of independent variable categories. The mean square estimate within the independent variable categories is calculated using a similar equation, Equation 9

[57].

SSW MS = (9) within N − k

where N is the number of observations within the groups. The mean square estimates are then divided by one another to calculate the F ratio which is a function of the amount of variation between the independent variables, see Equation 10 [57].

MS F = between (10) MS within

In order for the null hypothesis to hold true, the F value would be close to one. As the F value increases, the odds of rejecting the null hypothesis increases. The significance of the F value is found by comparing the Fcalculated value to the Fcritical value. The Fcritical value may be found in a reference table and is based on the significance level used (α=0.05 for this study) for the test as

59 well as the degrees of freedom previously mentioned. If the Fcalculated value is less than the Fcritical value, then one must fail to reject the null hypothesis [57]. Rather than finding the Fcritical value in a reference, the SAS software provides a p-value which is the probability of obtaining the test statistic value or a larger value if the null hypothesis was true. If the p-value is less than the alpha value, then the null hypothesis may be rejected and it may be concluded that there is a significant effect for the different denim samples. If the F value is considered significant, then there is a significant difference between the independent variable categories [56].

When the sample observations are unbalanced (i.e. unequal), then the least-squares means

(LSMEANS) may be used instead of the mean square estimate. The LSMEANS estimates what the marginal means would be if each sample had the same number of observations [56]. If the

MEANS statement in SAS was used instead of LSMEANS, the marginal means would be biased.

In order to determine which samples are significantly different, a Tukey Honestly Significantly

Different (Tukey HSD) test is performed on each of the pairwise comparisons [56]. It should be noted that a standardized student t-test would not be appropriate in this case, since a student t-test assumes that the two samples are randomly chosen. Following the ANOVA test the sample data is considered to be ordered and no longer random. The Tukey test accounts for difference, and a student t-test would be considered biased due to the different number of observations contributing to each mean.

The Tukey HSD test statistic, Q, may be calculated using the Equation 11 [58].

60 − XX Q = 21 (11) MSwithin N / sp

where the X variables are the means of the groups being compared and Np/s is number of observations within each group. If the two groups being compared have different numbers of observations, the Np/s may be calculated using Equation 12 [58].

2 N = (12) / sp ⎛ ⎞ ⎛ 11 ⎞ ⎜ ⎟ + ⎜ ⎟ ⎝ 1 ⎠ ⎝ NN 2 ⎠

where the N variables correspond to the number of observations in each of the groups being compared. The Q value is then compared to a tabulated Qcritical value. SAS also provides a p- value that may be used to determine if the Q value is significant. If the Q value is significant, then there is a significant difference between the two groups being compared.

2.4.2 Multivariate Analysis of Variance

Multivariate analysis of variance (MANOVA) is like the ANOVA analysis; however, multiple dependent variables may be used for the comparisons of the independent variable categories.

The MANOVA test “examines whether the population means on a set of dependent variables vary across levels of a factor or factors” [59]. The null hypothesis may also be stated as

61 ⎡M 11 ⎤ ⎡M 12 ⎤ ⎡M 1 j ⎤ ⎢ ⎥ ⎢M ⎥ ⎢M ⎥ M ⎢ 21 ⎥ ⎢ 22 ⎥ ⎢ 2 j ⎥ H 0 = ⎢M 31 ⎥ = ⎢M 32 ⎥ = ⎢M 3 j ⎥ (13) ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ MM ⎥ ⎢ M ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ M ⎥ ⎣M i1 ⎦ ⎣M i2 ⎦ ⎣ ij ⎦

where M represents a mean for one of the independent variable categories on one of the dependent variables [56]. The alternative hypothesis, Ha, would be that there is a significant difference between the independent variable categories when compared simultaneously across the dependent variables. Instead of finding the sum of squares, the MANOVA test uses the sum of squares and cross-products (SSCP) matrix, which is decomposed into within-groups, between- groups, and total SSCP matrices. Rather than finding a F test statistic, other test statistics are calculated, such as the Wilks’ Lambda test statistic, which is calculated using Equation 14 [59].

= TGW (14)

where G is the within-groups (error) SSCP matrix, T T is the total SSCP matrix, and indicate the determinant.

The MANOVA test uses the multivariate measure of association called Wilks’ lambda, which is a value between zero and one. A lambda value closer to zero indicates a relatively strong relationship between the independent variable categories (with replicate measurements taken as a group), and a lambda value closer to one indicates a relatively weak relationship. The F statistic that is used to test the significance of the relationship between the predictor (denim samples) and 62 multiple criterion variables (ion groups) is based on Wilks’ lambda [56]. When the p-value associated with the Wilks’ lambda is less than the threshold value of 0.05, the null hypothesis may be rejected and there is a significant multivariate effect for the samples. When the null hypothesis is rejected, it may be concluded that at least two of the independent variable categories are significantly different with respect to at least one dependent variable. In order to determine which dependent variable(s) resulted in the sample pairs being significantly different, both the previously mentioned Univariate ANOVA tests and Tukey HSD tests may be performed

[56]. All of these statistical tests may be performed using the SAS software. There are assumptions that are made when performing MANOVA with one between-subjects factor.

These typical assumptions include independent observations, random sampling, multivariate normality, and homogeneity of covariance matrices [56].

63 CHAPTER 3: OLIGOSACCHARIDE ANALYSIS

3.1 Instrument Parameters

The instrument used to conduct the analysis of the cotton samples and detect the oligosaccharides was a ThermoQuest Finnigan LCQDUO liquid chromatography (LC) instrument coupled with an ion trap mass spectrometer (MS) as the detector. An electrospray ionization

(ESI) source was used as the interface between the LC and MS. The instrument was tuned and calibrated according to the manufacturer specifications. The instrument was tuned, using a

+ maltopentaose (DP5) standard, to the 846 m/z peak corresponding to the [DP5+NH4] ion. A

TSP4000 LC pump was used to flush the mobile phase of 3mM ammonium acetate in 0.1% (v/v) acetonitrile buffer (pH ~4.8) through the system at a flow rate of 0.20 mL/min. The mobile phase was the source of hydrogen and ammonium ions for the formation of adducts.

In this research, sodium acetate and ammonium acetate buffers were examined as potential liquid chromatography mobile phases. The ammonium acetate produced a lower limit of detection, and a 3 mM concentration was chosen based on a paper by Mauri et al., which demonstrated that a cation concentration greater than 5 mM resulted in decreased ion abundance [35]. The acetonitrile was added to reduce the amount of bacteria that formed in the buffer solution, as was suggested by Thermo Finnigan. This small amount of acetonitrile did not affect the separation of the analytes; however, the acetonitrile had to be removed from the samples themselves by rotary evaporation in order for the samples to be separated chromatographically.

64

The MS acquire time was 20.00 minutes, with a divert value set to 0.20 minutes to send the mobile phase to the MS for 0.20 minutes then inject the sample and send the sample to the MS.

The MS was set to a full scan type of mass range 50.00-2000.00 m/z with a positive polarity and profile data type. The heated capillary temperature was set to 150ºC. Water blanks were analyzed between each standard/sample analysis. Xcalibur software was used to analyze the samples. The default ICIS (Interactive Chemical Information System) algorithm was used for auto-integration with the following default parameters: baseline window: 40, area noise factor 5, peak noise factor 10, no constraints on peak width, INCOS (Integrated Control System) noise algorithm method selected, minimum peak width: 3, multiplet resolution: 10, area tail extension: 5, and area scan window: 0. The mass-to-charge ratios listed in Table 3 were extracted over the full 20 minutes, which resulted in an extracted ion chromatogram.

Chromatographic separation was performed on a silica-based polar end-capped C18 reverse phase ThermoHypersil HyPURITY AQUASTAR column (250x2.1 mm, 5 μm, 0.35 mL/min column, part number 22505-252130) along with two guard columns (HyPURITY AQUASTAR,

5 μm, 10 mm x 2.1 mm, drop-in guard columns) [60].

HyPURITY AQUASTAR columns combine a C18 ligand with polar end-capping to produce a highly stable material for reversed phase chromatography. HyPURITY AQUASTAR columns have the unique ability to retain and separate polar analytes in 100% aqueous conditions, with no phase collapse. Selectivity and sensitivity can be maintained at decreased levels of buffer concentrations, making HyPURITY AQUASTAR ideal for applications involving MS detection [60].

65 The column was stored between uses in 1:1 water/acetone, per instruction by Thermo Finnigan.

Based on the dimensions of the column, a volume of 20 μL was injected onto the column, and the mobile phase flow rate was 0.2 mL/min [49]. In order to prevent contamination and inaccurate signal intensity due to carry-over of the analytes from one sample injection to another a single water blank was analyzed between samples.

Each sample was analyzed in triplicate using the LC-ESI-MS. The replicate analyses allowed for determination of the variance in the sample analyses due to the instrumentation. For each sample, Xcalibur software was used to find the oligosaccharide ion integrated peak areas, which is proportional to the number of ions detected. The auto-integration was manually adjusted when needed to exclude the peak corresponding to the DP1 ion group.

In order to be detected by the mass spectrometer, the analytes must be charged (positive or negative ions). In the experimental set-up, it was specified that this charge must be positive.

Three types of positively charged ions were utilized in this experiment. The first ion was the

[M+H]+ ion, which referred to the analyte (M) with a proton added to it. The second ion was the

+ [M+NH4] ion that resulted from the analyte forming an adduct with an ammonium ion. The final ion was the [M-OH]+ ion which occurred when a hydroxyl group (-OH) was eliminated from the analyte molecule leaving the molecule with a positive charge. The [M-OH]+ ion may also form when an oligosaccharide is cleaved to reduce the degrees of polymerization (Refer to

Figure 13).

66

Figure 13: Fragmentation of Oligosaccharide-Ammonium Adducts

However, only a minimal number of ions of this type were detected. These three types of ions may occur with analytes of different degrees of polymerization.

A degree of polymerization (DP) was defined in Equation 1 as the molecular weight of the polymer divided by the molecular weight of the repeating unit. The repeating unit for cellulose is cellobiose; however, for the purposes of this research the repeating unit used to calculate the

DP is the glucose residue. Degrees of polymerization greater than two are typically referred to as oligosaccharides. For the purpose of this research, degrees of polymerization (DP1-7) were referred to as oligosaccharides.

For the purposes of this research, analytes with a molecular weight of 180 amu such as dextrose and the glucose residues resulting from fragmentation of larger oligosaccharides as well as the

+ + + ions produced ([M+H] with m/z 181, [M+NH4] with m/z 198, and[M-OH] with m/z 163) will be abbreviated as DP1. Standard sucrose and extracted cellobiose with a molecular weight of

67 342 amu and ions will be abbreviated as DP2, maltotriose with a molecular weight of 504 amu and ions as DP3, maltotetraose with a molecular weight of 666 amu and ions as DP4, maltopentaose with a molecular weight of 828 amu and ions as DP5, maltohexaose with a molecular weight of 990 amu with ions as DP6, and maltoheptaose with a molecular weight of

1152 amu and ions as DP7. Table 3 summarizes the mass-to-charge ratio (m/z) for the ions formed for each degree of polymerization.

Table 3: Mass-to-Charge Ratios of the Ions/Adducts Corresponding to Oligosaccharides with Degrees of Polymerization 1-7 + + + Ion Group [M+H] [M+NH4] [M-OH] DP1 181 198 163 DP2 343 360 325 DP3 505 522 487 DP4 667 684 649 DP5 829 846 811 DP6 991 1008 973 DP7 1153 1170 1135

The Xcalibur software that was used with the instrumentation is capable of extracting specific ions for each sample analysis. For each sample, the three ions corresponding to each degree of polymerization (ion groups) were extracted and the total integrated peak area for the three ions was noted (Refer to Table 3 for list of mass-to-charge ratios extracted for each ion group). For example, after analyzing the sample, the 181, 198, and 163 m/z ions were extracted and the integrated peak area was found using the Xcalibur software. This integrated peak area was proportional to the amount of each DP1 ion type detected.

3.2 Oligosaccharide Extraction and Isolation from Cellulosic Material

68 3.2.1 Oligosaccharide Acid Extraction

Using the acid extraction method presented by Murray, the oligosaccharides were extracted from the denim [29]. One milliliter of 0.1 N hydrochloric acid was added to each 15 mL centrifuge tube containing the cellulosic material (or an oligosaccharide standard), and the tubes were placed in a hot water bath (100ºC) for thirty minutes. The oligosaccharides were released into the supernatant.

The acid extraction of oligosaccharides is less expensive and not as tedious as using a cellulase method for breaking down the cellulose into oligosaccharides. The other extraction method presented by Murray, using 0.5 mL of an acetic acid and nitric acid solution was also investigated [31]. Unlike the hydrochloric acid extraction which resulted in no visual change in the denim structure (aside from the color of the denim in certain cases), the acetic acid/nitric acid extraction resulted in a greater amount of degradation of the cotton fabric to a pulp-like state; however, oligosaccharides were not detected using the MS. (Procedural note: The SPE column was rinsed with 500 μL of 25% (v/v) acetonitrile and water.)

3.2.2 Isolation of Oligosaccharides

Graphitized carbon solid phase extraction has been shown to be an ideal method for the separation of sugars from dilute solutions containing salts, alkali, or mineral acid [39, 40]. The

SPE columns used in this work were Alltech Carbograph Solid Phase Extraction (SPE) columns

(150 mg bed weight, 4 mL volume size). The columns were conditioned immediately before use

69 by rinsing with three 1 mL aliquots of 80% (v/v) acetonitrile and water with 0.1% trifluoroacetic acid (TFA) followed by three 1 mL aliquots of water [40].

Following the hot water bath, 750 μL of supernatant was passed through a conditioned

Carbograph SPE column. The column was then rinsed with 2 mL of E-Pure water to remove the acid and glucose from the column. The oligosaccharides, DP2 and greater, were removed from the column by eluting five 1 mL aliquots of 25% (v/v) acetonitrile/water through the SPE column and collecting the eluent in a 25 mL round-bottom or pear-shaped flask.

Initial experiments revealed that the 25% (v/v) acetonitrile/water solution that was used to elute the analytes from the column diminished chromatographic separation on the reverse phase column, causing the oligosaccharides to elute at the same time. Therefore, the solutions were dried using a rotary evaporator and reconstituted in 750 μL of water prior to analysis using LC-

ESI-MS. (Note: For denim sample 12, extraction 2 on January 21, 2008, 650 μL was used instead of 750 μL, because less supernatant was available possible due to the absorption by the cotton fabric.).

Each denim sample was extracted in triplicate on each of two days, and the individual extractions were analyzed in triplicate on the LC-ESI-MS. Individual observations were excluded if the abundance of extracted oligosaccharides fell out of the linear dynamic range of the calibration curves.

70 3.3 Standards

3.3.1 Sample Collection and Preparation

An oligosaccharide kit containing maltotriose, maltotetraose, maltopentaose, maltohexaose, and maltoheptaose was purchased (Supelco, Bellefonte, PA, USA,) and stored in the dessicator.

Dextrose (Mallinckrodt Chemical Works, St. Louis, MO, USA) and sucrose (Aldrich Chemical

Company, Inc, Milwaukee, WI, USA) were purchased and stored at room temperature. A standard containing all of the mentioned oligosaccharides, dextrose, and sucrose was prepared at a concentration of approximately 100 ppm for each analyte. A 100 ppm DP7 standard was also prepared in order to determine the potential loss of oligosaccharides during the oligosaccharide extraction and isolation process.

3.3.2 Analyses and Results

3.3.2.1 Oligosaccharide Detection and Quantification

The ratio of extracted analytes with varying degrees of polymerization was unknown for the denim samples. If a mixture of analytes is to be ionized in ESI, some compounds may ionize more efficiently than others, which may result in the failure to detect some analytes due to competition for ion formation [52]. Separation of the analytes by liquid chromatography prior to ionization helps to alleviate this problem. By separating the analytes using liquid chromatography, the amount of each analyte with known degree of polymerization could be

71 determined by calibration, and dilutions made if necessary, to ensure analysis within the linear dynamic range (LDR) [52]. Initially, samples were analyzed using direct injection without liquid chromatography; however, due to competition for ion formation, it was determined that chromatographic separation of the analytes was needed.

A standard solution with known amounts of each analyte (glucose, sucrose, maltotriose, maltotetraose, maltopentaose, maltohexaose, and maltoheptaose) was separated by liquid chromatography and detected with ESI-MS. It was observed that a fraction of the analytes with higher degrees of polymerization fragmented to produce ions with lower degrees of polymerization, including the observation of ions corresponding to glucose. For this reason, the

+ + + [M+H] , [M+NH4] , and [M-OH] ions corresponding to all analytes (DP1-DP7, Refer to Table

3) were extracted to produce the EIC, in order to take into account the fragmentation. Refer to

Figure 14 for a sample EIC showing the integrated peak areas and Figure 15 for a corresponding

3-D map.

72 Figure 14: Extracted Ion Chromatograms for 10 ppm DP1-7 Standard

73

Figure 15: 3-D Map of Chromatogram for DP1-7 Standard

74 + It should be noted that occasionally small amounts of addition ions such as [2M+NH4] were

+ observed, primarily for DP2 and DP3 and [3M+NH4] ions were also observed for DP2.

[M+Acetate]+ ions were also observed on occasion. Although, the corresponding mass-to-charge ratios were not extracted when obtaining the integrated peak areas, the calibration curves produced were linear.

From dilutions of the mixture of approximately 100 ppm each DP1-7, calibration curves were constructed for each analyte. The analytes corresponding to DP4-7 each had a linear dynamic range of approximately 10.0 ng to 200 ng (or 0.500 ppm to 100 ppm) on the column. Due to the partial co-eluting of DP2 and DP3, the sum of the integrated peak areas for the two ion groups were used for the calibration curve; therefore, the linear dynamic range for DP2 and DP3 combined was approximately 20.0 ng to 400 ng (or 1.00 ppm to 200 ppm). The integrated peak area was determine to be within the LDR if the response fell within standard deviations greater than the upper end of the LDR (upper concentration) and two standard deviations below the lower end of the LDR (low concentration). A calibration curve was not constructed for DP1, since the monomer was not retained by the SPE column. A new calibration curve was constructed prior to each set of analyses to overcome instrument day-to-day variation.

3.3.2.2 Oligosaccharide Loss

In order to determine if the oligosaccharides were lost during the rotary evaporation process, a

100 ppm standard of maltoheptaose (DP7) was prepared and a 750 μL aliquot was placed in a conical flask. Five milliliters of 25% (v/v) acetonitrile and water was added to the flask, in order

75 to simulate the dilution that occurs when eluting the oligosaccharides from the SPE column. The flask was then rotary evaporated to dryness and the oligosaccharides reconstituted in 750 μL of water. The supernatant was then diluted 10:1 in water and was analyzed using LC-ESI-MS and a standard calibration curve prepared immediately before analysis of the sample.

Figure 16: Extracted Ion Chromatogram of DP7 Standard Following Rotary Evaporation

All of the maltoheptaose, within experimental error, was recovered, indicating that the oligosaccharides are not lost during rotary evaporation. Analysis of the base peak corresponding

+ to the analyte in the extracted ion chromatogram demonstrated that the [M+NH4] ion with m/z

1170 was present, and minimal fragmentation to shorter DP was observed in the spectrum.

To determine if the oligosaccharides were lost in the SPE column, another 750 μL aliquot of the

100 ppm DP7 was passed through a conditioned SPE column with the same extraction process previously described. The collected eluent was rotary evaporated to dryness and reconstituted in

76 750 μL of water. The supernatant was then diluted 10:1 in water and was analyzed using LC-

ESI-MS with a standard calibration curve prepared immediately before analysis of the sample.

Like the previous test results, the analyte did not appear to be lost on the SPE column or during rotary evaporation.

The same process was repeated following 500 μL of 100 ppm DP7 standard being boiled for 30 minutes with 500 μL of 0.2 N HCl, resulting in a final concentration of 50 ppm DP7 and 0.1 N

HCl. This simulated the acid extraction process that the denim, cotton t-shirt, and raw cotton samples would undergo, in order to determine if the oligosaccharides were broken down during the acid extraction process. In this case, the DP7 resulted in a 58% loss, and all of the smaller oligosaccharides were detected; therefore, the boiling the oligosaccharides in 0.1 N HCl results in an acid hydrolysis break-down of the larger oligosaccharides to the smaller oligosaccharides.

The total number of monosaccharides recovered was calculated, and a loss of 10% was observed following the acid wash; therefore, 90% of the moles of monosaccharides were recovered following hydrolysis. The loss may be accounted for by the formation of glucose monosaccharides during the acid wash that were then lost on the SPE column. The acid wash experiment was initially conducted with a standard containing DP1-7. The DP7 loss was calculated to be an average of 52%, and a greater amount of DP2 combined with DP3 was recovered than what was originally in the sample; therefore, as the DP7 standard test concluded, the larger oligosaccharides were breaking down to the smaller oligosaccharides. Since each sample was boiled in the same concentration and volume of acid for the same amount of time, the process was controlled and believed to have not introduced any additional error.

77

Figure 17: Extracted Ion Chromatogram of DP7 Standard Following Oligosaccharide Extraction Process

In addition, five milliliters of a 25 (v/v) ACN with 0.1% TFA solution was passed through the

SPE column, following oligosaccharide removal from the column by the protocol described above, in order to determine if “acidic sugars” were observed [40]. When the results were analyzed, no additional oligosaccharides were observed.

3.4 Denim

3.4.1 Sample Collection and Preparation

Denim samples manufactured with 100% cotton were purchased from the craft department at

Wal-Mart and from Jo-Ann Fabrics (Refer to Table 4). A section of each sample was cut into small pieces (ranging from approximately 3 mm and smaller) using scissors, and the pieces were

78 mixed until they were believed to be homogeneous with respect to their location within the original fabric section. Each sample (60.0 mg) was placed into a plastic 15 mL polypropylene conical centrifuge tube (17 mm x 120 mm) with screw-cap. For the sample size reduction test, additional 30.0 mg, and 15.0 mg of sample S8 were weighed in triplicate. In order to remove the variance in sample weight due to the wicking properties of cotton, all of the samples were stored in the same laboratory and weighed using the same analytical balance that has a precision of ±0.1 mg.

79 Table 4: 100% Cotton Denim Fabric Samples Purchased from Craft Sections Sample Purchase Purchase Store Item Manufacturing Description from Store Name Date Location* Code Country S1a 04/16/2007 A 8179786205 USA “Sportswear Denim by Oakhurst Textile”

S2 12/06/2006 B 256-3963 USA “Basic Denim Sunbleached Bottomweight”

S3a 04/16/2007 A 8179786079 Pakistan “Dress Denim by Oakhurst Textiles”

S4 12/06/2006 B 247-4740 China “Denim Basic Sunbleached Washed”

S5 12/06/2006 B 677-2933 Pakistan “Bottomweight Crosshatch Denim”

S6 12/06/2006 B 153-0245 China “Basic Denim, Indigo Wash Denim 04/16/2007 10oz”

S7 12/06/2006 B 157-3203 USA “Cotton Bottomweight Solid, Light 04/16/2007 Blue Ice Wash”

S8 04/16/2007 B 858-0417 China “Fashion Denim, 7.5oz Dark Blue CRSHT Denim Spring Bot”

S10 04/16/2007 B 836-9910 China “Fashion Denim, 8.2oz Drty htchd dnm bottomweight”

S11 04/16/2007 B 834-5621 Pakistan “Fashion Denim, Patriot BL BULL DNM, Bottomweight”

S12 04/16/2007 B 834-7189 China “Fashion Denim, MLTI EMB/EYELET BDR, Bottomweight (decoration is not 100% cotton)” S13 04/16/2007 B 858-0425 China “Fashion Denim 6.5oz Lt. BL SLUB DNM SPRING BOT”

S14 04/16/2007 B 858-0441 China “Fashion Denim, 7oz Blue Denim, Spring Bot”

*Purchase locations (purchased from craft/fabric sections of the store, not as prepared clothing): A = Wal-Mart Supercenter Store (11250 East Colonial Drive, Orlando, FL 32817) and B = Jo-Ann Fabric and Craft Superstore (825 North Alafaya Trail, Orlando, FL 32828-7049) Note: Sample S9 had the same barcode as sample S6 and sample S15 also had the same barcode as sample S7; therefore the samples with the same barcode were presumed to be the same and multiple purchase dates listed in the table. The samples were analyzed separately to ensure the results were consistent with one another.

80 3.4.2 Analyses and Results

3.4.2.1 Determination of Amounts of Oligosaccharides Extracted

Due to the mass-to-charge range limitation on the MS analyzer (m/z 150-2000), only those ion groups corresponding to DP1-10 could be detected by the instrument. Primarily DP2-7 were used for the analyses since DP1 was removed during the SPE column portion of extraction process [39], and the standards were only as large as DP7. DP8-DP10 standards were not used to establish calibration curves. However, these larger oligosaccharides were observed, but the areas detected from the samples were significantly less than for the other oligosaccharides. It was observed that between 3.5 and 4 minutes in the chromatograms, a portion of what was referred to as the DP2 and DP3 extracted ion chromatogram peak contained all of the oligosaccharide ions. It is unclear as to why this small amount of oligosaccharides was not separated chromatographically with the others, but it should be noted that this may be a source of instrumental error. One possibility to explain this phenomenon would be that the analytes observed between 3.5 and 4 minutes have the same molecular weights as the other oligosaccharides, but different structures which would affect their chromatographic separation.

Alternatively, there may be a small amount of oligosaccharides eluting in the column excluded volume. Small retention time shifts were observed, but ion group peaks were confirmed by mass spectral analyses. A sample extracted ion chromatogram for each of the denim fabric samples may be observed in Figure 18, and corresponding 3-D maps for denim samples may be observed in Figure 19. All of the denim samples produced similar results, with the exception of denim sample S2 that appeared to have a decreased amount of the DP6 and DP7 ion groups. An initial

81 extraction of each denim material was performed in order to determine the dilution factor needed to analyze the denim samples within the linear dynamic range.

Figure 18: Extracted Ion Chromatograms for Each Denim Sample

82

Figure 19: 3-D Map of Chromatograms Denim Samples S6 and S2

83 A calibration curve was produced using the oligosaccharide standards prior to each set of sample analyses. Refer to Figure 20 for sample calibration curves with y-axis error bars showing one standard deviation above and one standard deviation below the mean response values and Table

5 for LDR equations and corresponding R2 values that are based on the mean response values for each concentration.

84

Figure 20: Sample Standard Calibration Curves for Each Ion Group

85 Table 5: Equations for Denim Sample Calibration Curves and Respective R2 Values Samples Date Extracted DP2&3 DP4 DP5 DP6 DP7 (combined) 3a, 6, 10 20071016 y = 4.90E+05x + y = 1.33E+06x + y = 1.32E+06x + y = 1.68E+06x + y = 1.34E+06x + 4.82E+06 2.93E+06 6.29E+06 1.40E+06 3.11E+06 0.999 1.00 1.00 1.00 1.00 20071025 y = 5.64E+05x + y = 1.67E+06x + y = 1.62E+06x + y = 1.98E+06x + y = 1.66E+06x + 1.41E+07 1.27E+07 8.17E+06 8.90E+06 4.77E+06 0.995 0.998 1.00 1.00 1.00 4, 15 20071106 y = 5.54E+05x + y = 1.75E+06x + y = 1.64E+06x + y = 1.92E+06x + y = 1.50E+06x + 1.08E+07 5.06E+06 5.38E+06 5.37E+06 4.69E+06 0.997 1.00 1.00 1.00 1.00 20071113 y = 4.00E+05x - y = 1.45E+06x + y = 1.43E+06x + y = 1.71E+06x + y = 1.44E+06x + 6.61E+05 2.40E+06 1.79E+05 2.07E+06 2.24E+06 0.989 1.00 1.00 1.00 1.00 5, 8, 9 20071207 y = 5.21E+05x + y = 1.45E+06x + y = 1.51E+06x + y = 1.70E+06x + y = 1.39E+06x + 6.91E+06 2.87E+06 5.52E+05 9.95E+03 9.02E+05 0.999 1.00 0.999 0.999 0.999 20071211 y = 1.21E+06x + y = 2.89E+06x + y = 2.74E+06x + y = 2.75E+06x + y = 2.32E+06x + 2.91E+07 7.62E+06 2.06E+07 1.79E+07 1.28E+07 0.986 1.00 0.994 0.993 0.997 13, 14 20071218 y = 3.97E+05x + y = 1.16E+06x + y = 1.50E+06x - y = 1.67E+06x - y = 1.38E+06x - 5.64E+06 4.97E+05 1.04E+06 9.88E+06 2.79E+06 1.00 0.999 0.999 0.998 0.996 20080109 y = 4.49E+05x - y = 1.21E+06x - y = 1.49E+06x - y = 1.59E+06x - y = 1.39E+06x - 1.47E+06 7.42E+06 7.35E+06 1.04E+07 8.67E+06 0.988 0.993 0.996 0.993 0.993 1a, 11, 12 20080118 y = 8.22E+05x + y = 1.71E+06x + y = 1.79E+06x + y = 1.74E+06x - y = 1.46E+06x + 6.54E+06 1.10E+07 5.33E+06 2.16E+06 2.80E+06 0.999 0.999 0.999 1.00 1.00 20080122 y = 8.42E+05x + y = 1.68E+06x + y = 1.96E+06x + y = 1.93E+06x + y = 1.80E+06x + 1.50E+07 7.28E+06 1.21E+07 9.48E+06 3.72E+06 0.997 1.00 0.996 0.999 1.00 2, 7 20080129 y = 7.29E+05x + y = 1.27E+06x + y = 1.70E+06x + y = 1.70E+06x + y = 1.39E+06x + 7.56E+06 1.16E+07 6.30E+06 3.29E+06 5.63E+06 1.00 1.00 1.00 0.999 1.00 20080201 y = 7.98E+05x + y = 1.42E+06x - y = 1.88E+06x + y = 1.85E+06x + y = 1.41E+06x + 7.57E+06 6.14E+06 6.79E+06 6.03E+06 1.23E+07 1.00 0.963 1.00 1.00 0.995

86 The amount of recovered oligosaccharides following the denim acid extraction process was determined based on the standard calibration curves. The ion abundance was measured using the same method as the standard calibration curves. The dilution factor of the solutions was taken into account, and an absolute value of recovered extracted oligosaccharides was recorded. These absolute values were placed in a matrix and analyzed using SAS software. All of the data that was within the linear dynamic range from the two extraction days were used (Refer to Table 6 for a list of the number of observations used for each sample and Appendix A for the SAS code including a list of all data).

87 Table 6: Number of Extractions and Observations within the Linear Dynamic Range of the Calibration Curve for the Denim Samples (Including Sample Repeats S9 and S15) Sample Extraction Day Extractions Observations Total Performed Within LDR Observations* S1a 1 3 9 18 2 3 9 S02 1 3 9 18 2 3 9 S3a 1 2 6 15 2 3 9 S04 1 3 9 18 2 3 9 S05 1 3 9 18 2 3 9 S06 1 2 6 15 2 3 9 S07 1 3 9 18 2 3 9 S08 1 3 9 18 2 3 9 S09 1 2 6 12 2 2 6 S10 1 2 6 15 2 3 9 S11 1 3 9 18 2 3 9 S12 1 3 9 18 2 3 9 S13 1 3 9 18 2 3 9 S14 1 3 9 18 2 3 9 S15 1 3 9 18 2 3 9

88

Figure 21: 3-D Map of Chromatograms for a DP1-7 Standard and Denim Sample S5

89 It was observed that the DP2 standard (sucrose) eluted at a slightly different time than the DP2 from the samples. This occurred because sucrose is used as the standard and not maltose.

Although the two molecules have the same molecular weight, sucrose is composed of both a six and five membered ring (glucose plus fructose) and maltose is composed of two six membered rings (glucose and glucose). The position change in the hydroxyl group affects the way the analyte is retained, which was observed by Lui et al. [34].

3.4.2.2 Comparison of Thirteen Denim Samples

3.4.2.2.1 Comparison Using the Different Ion Groups

When collected data consists of a single independent variable measured on a nominal scale (i.e. fabric samples) and multiple criterion variables measured on an interval or ratio scale (i.e. ion groups, based on the degrees of polymerization of the analytes), multivariate analysis of variance

(MANOVA) with one between-subjects factor may be used to statistically analyze the data [56].

The SAS software was used to analyze the data using the MANOVA test followed by individual

Univariate ANOVA tests, and Tukey Honestly Significantly Different (Tukey HSD) tests. The

SAS code used for the analyses may be found in Appendix A.

When analyzing the results from a MANOVA test, one must first consider the Wilks’ Lambda F statistic (4.22x10-4) and the corresponding p-value (<0.0001) which are used in deciding whether or not to accept the null hypothesis that there is no significant difference between the samples when compared simultaneously across all of the following five ion groups: DP2 and DP3

90 (combined), DP4, DP5, DP6, and DP7. In this case, since the Wilks’ Lambda F statistic is significant because the p-value is less than the alpha value 0.05, one would conclude that there is a significant multivariate effect for the ion groups. In other words, at least two of the samples are significantly different with respect to at least one ion group; therefore, one may reject the null hypothesis that there is no difference in the denim fabric samples when compared simultaneously on the oligosaccharide ion groups DP2-7 extracted. Therefore, the results from the individual

ANOVA tests are considered.

When the null hypothesis is rejected, it may be concluded that at least two of the samples are significantly different with respect to at least one ion group. In order to determine which ion group(s) result in the sample pairs being significantly different, both Univariate ANOVA tests and Tukey HSD tests may be performed [56].

The null hypothesis of the Univariate ANOVA test may be stated as: In the population, there is no difference between the denim fabric samples with respect to the mean amount, in nanograms, of oligosaccharides of a certain ion group (a particular degree of polymerization). For example, the mean amount of DP4 (in nanograms) detected for denim sample S1a is the same for denim sample S2, and S3, etc.

Since the entire set of sample observations were not used, due to values falling outside of the linear dynamic range, the number of observations for each sample was not equal; therefore, the

LSMEANS statement had to be used to calculate the marginal means. The results of the individual univariate (ANOVA) tests were tabulated in Table 7.

91

Table 7: Results of Individual Univariate Analyses on the Ion Grouping Criterion Values Ion Group F Value P-value DP2 and DP3 855 <0.0001 DP4 303 <0.0001 DP5 512 <0.0001 DP6 248 <0.0001 DP7 207 <0.0001

For all of the ion groups, the F values are considered to be statistically significant since the p- values are less than the alpha value of 0.05; therefore, the null hypothesis of no population differences is rejected and one can conclude that there is a significant difference between the denim fabric samples. Since the null hypotheses were rejected in all cases, the results for the

Tukey HSD tests may then be analyzed for all ion groups: DP2 with DP3, DP4, DP5, DP6, and

DP7. The Tukey HSD tests for each ion group were interpreted and the results summarized in

Figure 22.

92

Figure 22: Tukey HSD Results of Pairwise Comparisons

Thirteen denim samples were used; therefore seventy-eight pairwise comparisons were calculated. A “1” in a cell indicates that the samples were significantly different from one another, and a “0” in a cell (also highlighted in yellow) indicates that the two denim samples were not significantly different from one another. The percent discriminated values were calculated using equation 7. 93

[()Number − wisePair Comparisons − (Number Significan Differenttly Sample Pairs)] ×100 (7) ()Number − wisePair Comparisons

The comparisons for all of the ion groups were combined, with the cells highlighted in orange representing a pairwise comparison that was discriminated in at least one of the ion groups. The remaining yellow cells were not discriminated by any of the ion groups. The overall pairwise discrimination of the thirteen denim samples using all five ion groups was 85.9%. Table 8 summarizes the results.

Table 8: Summary of Tukey HSD Results and Calculated Percent Discrimination Ion Group # Pairs Sig. Different % Discrimination DP2 and DP3 66 84.6 DP4 64 82.1 DP5 63 80.8 DP6 64 82.1 DP7 64 82.1 Combined DP2 through DP7 67 85.9

The alpha value corresponds to the probability of a Type I error occurring, where a Type I error means incorrectly stating that the means of the two different samples are significantly different.

With 85.9% of the denim samples analyzed with my specific test are significantly different from one another. The power of the test (1-Beta) is therefore 85.9%, which was determined using known sample comparisons. Therefore, the probability of committing a Type II error, which would be stating that the means of two samples are not significantly different when in fact they are, is equal to 1-0.859 = 0.141. There is a 14.1% probability of committing a Type II error.

94 It is important to note that these results operate under the assumption that all of the samples purchased with different bar codes are in fact different. [Note: The S6 and S7 samples were purchased on one day, then samples with the same barcode were purchased on a second day

(refer to Table 4) and originally referred to as S9 and S15 respectively. All four samples were extracted in triplicate and analyzed on the LC-ESI-MS in triplicate. Only those observations falling within the linear dynamic range were kept for analysis. An initial MANOVA/Univariate

ANOVA/Tukey HSD analysis was conducted by giving the four samples different names, and the results demonstrated that the S6 and S9 sample were not significantly different from one another across any of the ion groups. The same results occurred for the S7 and S15 samples.

Therefore, the S6 and S9 samples were combined and given only the name S6, and the S7 and

S15 samples were combined and given the name S7.]

3.4.2.2.2 Comparison Using the Total Moles of Monosaccharides Calculated

The same data set from the previous analysis was statistically analyzed using a second method.

The overall monosaccharide amount was calculated for each sample by taking the nanograms detected, converting to grams, then multiplying by the molecular weight of the analyte and multiplying by the number of monosaccharide units in that particular analyte (Refer to Equations

14 and 15 for examples). For the DP2 and DP3 combined nanograms detected, the average mass was used [(342 + 504) / 2 = 423 amu] as well as the average monomers [(2 + 3) / 2 = 2.5 monomer units].

95 ⎡ ⎛ DPg 3&21 ⎞ ⎛ DPmole 3&21 ⎞ ⎛ 5.2 molesMonosaccharides⎞⎤ DPng 3&2 ×⎜ ⎟×⎜ ⎟×⎜ ⎟ (14) ⎢()⎜ 9 ⎟ ⎜ ⎟ ⎜ ⎟⎥ ⎣ ⎝ × DPng 3&2101 ⎠ ⎝ DPg 3&2423 ⎠ ⎝ DPmole 3&21 ⎠⎦

⎡ ⎛ DPg 51 ⎞ ⎛ DPmole 51 ⎞ ⎛ 5 moles Monosaccharides ⎞⎤ DPng 5 × ⎜ ⎟ × ⎜ ⎟ × ⎜ ⎟ (15) ⎢()⎜ 9 ⎟ ⎜ ⎟ ⎜ ⎟⎥ ⎣ ⎝ ×101 DPng 5 ⎠ ⎝ DPg 5828 ⎠ ⎝ DPmole 51 ⎠⎦

The amount of a single monomer unit for all of the different ion groups was calculated and summed to provide a total amount of monomer units detected for each sample (Refer to

Appendix B).

Since only one dependent variable was used for the comparison of the independent variable categories, the Univariate ANOVA test was used. When analyzing the results from a Univariate

ANOVA test, one must first consider the F value (457) and the corresponding p-value (<0.0001).

The p-value is less than the alpha value 0.05; therefore, the null hypothesis of no population differences is rejected and one can conclude that there is a significant effect for the denim fabric samples. Since the null hypotheses was rejected, the results for the Tukey HSD test may then be analyzed.

The Tukey HSD test was interpreted and the results summarized in Figure 23.

96

Figure 23: Tukey HSD Results of Pairwise Comparisons Using Total Moles of Monosaccharides

The cells containing a “0” (yellow cells) did not have least-squares means that were significantly different from one another, and the overall pairwise discrimination of the thirteen denim samples was 82.0%. The alpha value corresponds to the probability of a Type I error occurring, where a

Type I error means incorrectly stating that the means of the two different samples are significantly different. With the Tukey HSD procedure in MANOVA (using LSMEANS), at a nominal α level at 0.05, 82.0% of the denim samples analyzed with my specific test are significantly different from one another. The power of the test (1-β) is therefore 82.0%, which was determined using known sample comparisons. Therefore, the probability of committing a

Type II error, which would be stating that the means of two samples are not significantly different when in fact they are, is equal to 1-0.820 = 0.180. There is a 18.0% probability of committing a Type II error.

97 As was previously stated, it is important to note that these results operate under the assumption that all of the samples purchased with different bar codes are in fact different. An initial

Univariate ANOVA/Tukey HSD analysis was conducted by giving the four samples different names, and the results demonstrated that the S6 and S9 sample were not significantly different from one another across any of the ion groups. The same results occurred for the S7 and S15 samples. Therefore, the S6 and S9 samples were combined and given only the name S6, and the

S7 and S15 samples were combined and given the name S7.

3.4.2.3 Sample Size Reduction Test

In order to investigate using the method for forensic size samples (single fiber), a sample size reduction test was performed. For the sample size reduction test, additional 30.0 mg and 15.0 mg of sample S8 were extracted in triplicate and analyzed in triplicate, like the other denim samples. The amount of oligosaccharides (ng) and total monosaccharides extracted were also calculated and combined with the results from the previous 60.0 mg S8 denim sample analysis

(Refer to Table 9).

Table 9: Amount (ng) of Each Ion Group Detected and the Total Moles of Monosaccharides Extracted for Sample Size Reduction Test Weight of DP2&3 DP4 (ng) DP5 (ng) DP6 (ng) DP7 (ng) Total Moles Denim (ng) Monosaccharides Sample Extracted (mg) Average* 60.0 1029 958 885 826 839 2.73E-08 30.0 594 314 332 245 372 1.11E-08 15.0 437 324 350 329 369 1.09E-08 Standard Deviation (Percent Relative Standard Deviation)*

98 60.0 225 (21.9) 168 (17.5) 165 (18.6) 167 (20.2) 161 (19.2) 6.28E-09 (23.0) 30.0 32 (5.39) 40 (12.7) 28 (8.43) 26 (10.6) 44 (11.8) 3.39E-10 (3.05) 15.0 29 (6.64) 46 (14.2) 19 (5.43) 27 (8.21) 26 (7.05) 5.62E-10 (5.16) * The number of observations used to calculate the 60 mg, 30 mg and 15 mg denim sample sizes were n=18, n=9, and n=9, respectively.

The results showed that the extracted amount of material was proportional to the amount of denim sample used when reducing from 60.0 mg to 30.0 mg, but it was not proportional to the amount of denim sample when reduced from 30.0 mg to 15.0 mg. Based on these results, it would be possible to reduce the sample size to 30.0 mg, but further reduction would require method modifications.

3.5 White Cotton T-Shirts

3.5.1 Sample Collection and Preparation

White cotton t-shirt samples manufactured with 100% cotton were purchased from the Men’s department at Wal-Mart (Refer to Table 10). The samples were prepared the same way as the denim samples (Refer to 3.4.1 Sample Collection and Preparation).

99

Table 10: White 100% Cotton T-Shirt Samples Sample Purchase Store Item Manufacturing Description Date* Code Company F 09/19/2007 007603127273 El Salvador Fruit of the Loom Style 2727

H 09/19/2007 007533800361 Honduras Hanes Style 2135

J 09/19/2007 078937444197 Honduras Jockey

P 09/19/2007 076615955499 Malaysia Puritan 100% Ringspun cotton

*Purchase location: Wal-Mart Supercenter Store (11250 East Colonial Drive, Orlando, FL 32817)

3.5.2 Analyses and Results

The white cotton shirts were analyzed using the same method as the denim samples; however, only two extractions were performed. The first extraction was analyzed with the LC-ESI-MS in duplicate, and the second sample was only analyzed once. The oligosaccharide ion groups (DP2 and DP3 through DP7) were analyzed like the denim samples, following calibration. The extracted ion group with a retention time similar to the DP7 ion group of the standard exhibited a mass-to-charge ratio that was the same as the DP4 ion group of the standard. Like the denim samples, all of the oligosaccharides were observed in the first half a minute of the peak with the same retention time of the DP2 and DP3 standard peak. The [M+H]+ has a greater abundance for the white cotton t-shirts than for the denim. All of the samples had similar extracted ion chromatograms; however, for the first J sample [J (1)] small amounts of each of the ion groups were observed at the expected retention times, but this did not occur for the second J extraction

[J (2)] (See Figure 24 for the Extracted Ion Chromatograms and Figure 25 for the corresponding

3-D Maps for the first and second J sample extracts).

100

Figure 24: Extracted Ion Chromatograms for White Cotton T-Shirt Samples

101

Figure 25: 3-D Map of Chromatograms for a White Cotton T-Shirt Sample J, Extractions 1 and 2

102 3.6 Raw Cotton

3.6.1 Sample Collection and Preparation

Raw cotton samples were obtained from The International Cotton Association Limited (ICA

Ltd.) in Liverpool, UK. The samples provided by ICA, Ltd. were grown in Tajikistan,

Uzbekistan, Egypt, Iran, and Benin West Africa (For HVI data, refer to Table 11) [61]. Initially, the raw cotton samples were not washed prior to being weighed, but impurities in the sample were removed.

Table 11: HVI Data for ICA Ltd. Samples Sample Country/Region Mic Len Unf Str Elg Rd +b TrCt R1 Tajikistan 4.47 1.115 82.5 28.8 6.5 77.2 10.03 7 R2 Uzbekistan 4.47 1.115 82.5 28.8 6.5 77.2 10.03 7 R3 Egypt 4.45 1.306 85.5 38.5 7.3 72 11.7 8 R4 Iran 3.99 1.377 87. 42.7 6.6 67 12.2 17 R5 Benin West 4.39 1.318 85.1 37.9 7.2 73 11 5 Africa Mic = Micronaire (width of the fiber), Len = Length of the cotton fibers (inches), Unf = Percentage of fibers with equal breaking strength, Str = Breaking point, Elg = Stretching point until the fibers break, Rd = Color – white, +b = Color – yellow/grey, TrCt = amount of trash particles found in sample. Samples received by the National Center for Forensic Science on January 17, 2008.

Another set of raw cotton samples was washed using the cold water extraction described by

Murray [31]. The samples were weighed (60.0 mg) and placed in individual centrifuge tubes, and one milliliter of E-pure water was added to each tube. The tubes were sonicated in an ice water bath for thirty minutes, the supernatant removed, and the samples were dried in an oven at

66°C over-night.

103

3.6.2 Analyses and Results

Both sets of samples were analyzed once using the LC-ESI-MS. The cold water wash did not appear to make a significant difference in the results. The results were similar to the white t- shirts in that the retention times and mass-to-charge ratios were not the same as the standard samples and all of the oligosaccharide ions were observed in the first portion of the DP2 and

DP3 extracted ion peak. Unlike the white cotton t-shirts, the DP2 and DP3 peak had an integrated peak area that corresponded to a concentration greater than the linear dynamic range and small amounts of the DP7 ions were observed, but the molecular ions for the peak with the retention time similar to standard DP7 was still consistent with DP4 ions. If the sample solutions were diluted so the DP2 and DP3 ion groups fall within the linear dynamic range, the remaining extracted ion peaks (DP4-7) would have ion abundances below the linear dynamic range. Refer to Figure 26 for the extracted ion chromatograms for each of the raw cotton samples as well as

Figure 27 for the corresponding 3-D Map of raw cotton sample R1 as well as a standard DP1-7 sample for comparison.

104

Figure 26: Extracted Ion Chromatograms for Raw Cotton Samples

105

Figure 27: 3-D Map of Chromatograms for a DP1-7 Standard and Raw Cotton Sample R1

106 CHAPTER 4: DYE ANALYSIS

4.1 Instrumentation Parameters

A Thermo GC2000 gas chromatograph coupled with a PolarisQ mass spectrometer was used for the analysis of the dye extracted from the denim fabrics. The samples were chromatographed on a 12 m methyl siloxane capillary column with a 0.2 mm ID and 0.33 μm film thickness. Samples were injected (splitless) into the 200°C injector port at a sample volume of approximately 1.0

μL. The initial column temperature was 60°C and was ramped at 10°C/min to 185°C then ramped 25°C/min to 250°C and held for five minutes. Spectra were scanned over a range of m/z

40-1000, and the results examined using the extracted ion chromatograms for the extracted ion with m/z 262.

4.2 Standards

An indigo dye standard was prepared by dissolving a synthetic indigo standard (Aldrich

Chemical Company, Inc, Milwaukee, WI, USA) in chloroform.

4.2.1 Sample Collection and Preparation

The indigo dye from the denim samples was extracted from a 10 mm blue thread using approximately 20 μL of chloroform and heating at 100°C for twenty minutes. This method was

107 previously shown to completely extract indigo dye from cotton and viscose samples [62]. The extraction was accomplished by placing the thread in a glass melting point capillary tube, adding the chloroform with a syringe, then dipping the tube in liquid nitrogen in order to freeze the chloroform. Once the chloroform was frozen, a torch was used to seal the open end of the capillary tube. The tube was then heated in the oven. The supernatant was injected into the GC-

MS.

4.2.2 Analyses and Results

The standard was eluted at approximately sixteen minutes (See Figure 28 for the extracted ion chromatogram and mass spectrum corresponding to the peak in the EIC).

108

Figure 28: Extracted Ion Chromatogram and Spectrum for Indigo Standard

The extracted ion chromatograms and spectra were examined for the denim samples. Figure 29 shows the EIC and spectrum for sample S3a.

109

Figure 29: Extracted Ion Chromatogram and Spectrum for Denim Sample S3a

Indigo was found in all of the samples except S11. Isatin (Figure 30), formed when indigo is oxidized with nitric acid or chromic acid [63], eluted from the column at 8.25 minutes and was observed in the following samples: S1, S9, S12, S13, S14, S15 (Recall that S15 is the same as

S7 and S9 is the same as S6, but the samples were purchased on a different days).

Figure 30: Isatin

110 Isatin was not observed at a signal-to-noise of 3:1 for S7, but was for S15, and it was not observed in S6; therefore, the isatin may not be homogeneously dispersed across the denim fabric and was not used to aid in discrimination of the denim fabric samples.

111 CHAPTER 5: DISCUSSION

5.1 Current Research

When analyzing the thirteen denim samples, the relative amounts of the oligosaccharides extracted appeared to be unique for the different denim samples, which is in agreement with results by Murray who concluded that the relative concentrations of the oligomers extracted were characteristic of the source of the cotton, but Murray did not show a statistically valid discrimination based on quantity of oligosaccharides and did not use mass spectrometry to accurately identify each analyte and correct for co-eluting impurities [32]. However, the sample reduction size test with the denim samples demonstrated that the amount of oligosaccharides extracted was not proportional to the amount of denim fabric used below 30.0 mg of sample.

When comparing the two statistical methods, by using the amount (ng) of each of the oligosaccharide ion groups extracted or the total moles of monosaccharide groups extracted, the results were similar.

112

Figure 31: Comparison of Tukey HSD Results for Both Statistical Methods

In both of the methods, the pairwise comparisons with a “1” (white cell), in Figure 31, were significantly different, and the pairwise comparisons with “0” (highlighted in yellow) were not considered significantly different using either method. Those sample comparisons with a “1”

(highlighted in orange) were found to be statistically different when the oligosaccharide ion groups were considered separately, but not when the total moles of monosaccharide were considered. Sample comparisons S4 and S12 as well as S7 and S12 were considered to be statistically different based on the amount (ng) of DP2 and DP3 ion groups present in the samples. Samples S11 and S12 were statistically different when looking at either the amount

(ng) of DP6 ion groups or DP7. However, since all of the ion groups were observed in a portion of the DP2 and DP3 extraction ion chromatogram peak, this may have introduced error into the method.

113 When combining the results from the denim dye analysis with the denim oligosaccharide extraction results, sample S11 was able to be discriminated from samples S5 and S6 (pink highlighted cells), since indigo dye was not present in sample S11 but was for both S5 and S6.

Figure 32: Discrimination Matrix Based on Different Ion Groups and the Presence of Indigo Dye

The overall discrimination of the thirteen denim sample was then increased to 88.5%; therefore, when the two methods were combined, the probability of committing a type II error was 11.5%

(Refer to Figure 32).

When combining the results from the denim dye analysis with the total monosaccharides detected from the denim extraction results, sample S11 was able to be discriminated from samples S5, S6, and S12 since indigo dye was not present in sample S11 but was for S5, S6, and

S12. The overall discrimination was then increased to 85.9%, with the probability of committing a type II error equal to 14.1% (Refer to Figure 33).

114

Figure 33: Discrimination Matrix Based on the Total Moles of Monosaccharides and the Presence of Indigo Dye

Additional cellulosic materials were also investigated including four white 100% cotton t-shirts as well as five raw cotton samples grown in Tajikistan, Uzbekistan, Egypt, Iran, and Benin West

Africa. The analytical methodology gave results for the white cotton t-shirts and raw cotton samples that were inconsistent with those obtained from the denim samples. Both samples produced ions with the same mass-to-charge ratios as the standards and denim samples; however, the retention times of the ions were not consistent. This result could indicate that the analytes from the white cotton t-shirts and raw cotton were oligosaccharides of different molecular structure, resulting in different retention times.

5.2 Future Research

Future research that may be conducted would involve the analyses of other cotton textiles such as bath towels or other cellulose-containing samples such as wood and paper products and compared to the results of Murray [9]. The results of the analysis may also be compared with the

115 instrumentation that Murray used, a HPAEC-PAD [32]. Other types of denim materials may be analyzed as well include those of “stretch denim” which contain spandex [19]. Dye analyses may be performed by looking for dyes other than indigo, such as indigo derivatives [20]. This may add to the discrimination of the denim samples. Additional analysis of the fibers using UV-

Visible absorption spectroscopy, like Suzuki et al. may also aid in the discrimination of the denim samples [47]. More work on sample size reduction to accommodate forensic-size (single fiber) samples should also be investigated.

116 APPENDIX A: SAS CODE FOR DENIM SAMPLES USING AMOUNT OF OLIGOSACCHARIDES (NANOGRAMS) FOR DIFFERENT ION GROUPS

117 Data Frisch_20080215_Method1;

Input #1 @1 Sample $3.

@5 DP23 6.

@12 DP4 6.

@19 DP5 6.

@26 DP6 6.

@33 DP7 6.;

Datalines;

S1a 5.82e3 2.50e3 3.01e3 2.44e3 2.34e3

S1a 5.69e3 2.65e3 2.45e3 2.67e3 2.66e3

S1a 5.89e3 2.68e3 2.55e3 2.38e3 2.83e3

S1a 2.90e3 1.63e3 1.88e3 1.86e3 2.08e3

S1a 2.52e3 1.41e3 1.63e3 1.96e3 1.89e3

S1a 2.98e3 1.81e3 1.89e3 1.93e3 1.97e3

S1a 4.68e3 2.05e3 2.05e3 2.02e3 2.07e3

S1a 4.65e3 2.28e3 2.18e3 2.06e3 2.21e3

S1a 5.11e3 2.13e3 2.32e3 2.52e3 2.32e3

S1a 3.70e3 1.71e3 1.83e3 1.58e3 1.55e3

S1a 3.59e3 1.68e3 1.68e3 1.44e3 1.53e3

S1a 3.77e3 1.73e3 1.78e3 1.61e3 1.76e3

S1a 5.22e3 2.56e3 2.74e3 2.33e3 2.29e3

S1a 5.25e3 2.61e3 2.47e3 2.42e3 2.36e3

S1a 5.04e3 2.52e3 2.51e3 2.36e3 2.63e3

118 S1a 5.42e3 2.59e3 2.62e3 2.31e3 2.29e3

S1a 5.46e3 2.66e3 2.59e3 2.45e3 2.27e3

S1a 5.33e3 2.75e3 2.47e3 2.36e3 2.19e3

S02 1.32e4 2.96e3 5.53e3 9.56e2 1.14e3

S02 1.25e4 2.83e3 4.14e3 7.68e2 1.21e3

S02 1.25e4 3.06e3 4.81e3 8.65e2 1.07e3

S02 1.21e4 3.89e3 5.48e3 8.43e2 6.62e2

S02 1.17e4 2.80e3 4.63e3 6.28e2 9.06e2

S02 1.28e4 3.37e3 4.50e3 7.15e2 8.31e2

S02 1.56e4 4.78e3 5.79e3 9.47e2 1.05e3

S02 1.30e4 4.77e3 6.53e3 1.25e3 1.17e3

S02 1.37e4 4.07e3 4.99e3 7.83e2 1.01e3

S02 1.41e4 4.55e3 4.65e3 7.26e2 7.72e2

S02 1.35e4 3.09e3 3.66e3 5.85e2 6.45e2

S02 1.29e4 3.50e3 5.25e3 9.08e2 8.03e2

S02 1.35e4 3.63e3 4.61e3 6.93e2 4.67e2

S02 1.40e4 3.99e3 4.52e3 7.17e2 3.51e2

S02 1.55e4 4.33e3 5.45e3 1.03e3 3.96e2

S02 1.50e4 5.05e3 5.25e3 1.08e3 8.70e2

S02 1.30e4 3.47e3 4.18e3 5.40e2 4.11e2

S02 1.40e4 3.77e3 4.37e3 6.46e2 4.69e2

S3a 2.34e3 1.24e3 1.35e3 1.21e3 1.23e3

S3a 2.45e3 1.23e3 1.33e3 1.17e3 1.24e3

119 S3a 2.56e3 1.33e3 1.42e3 1.26e3 1.27e3

S3a 3.15e3 1.55e3 1.59e3 1.45e3 1.58e3

S3a 3.03e3 1.55e3 1.48e3 1.32e3 1.39e3

S3a 2.84e3 1.44e3 1.58e3 1.38e3 1.36e3

S3a 3.75e3 1.77e3 1.88e3 1.64e3 1.68e3

S3a 3.23e3 1.72e3 1.86e3 1.60e3 1.62e3

S3a 3.06e3 1.67e3 1.73e3 1.47e3 1.49e3

S3a 2.78e3 1.64e3 1.65e3 1.46e3 1.42e3

S3a 2.75e3 1.40e3 1.48e3 1.30e3 1.28e3

S3a 2.36e3 1.40e3 1.55e3 1.44e3 1.40e3

S3a 2.83e3 1.63e3 1.59e3 1.40e3 1.35e3

S3a 2.41e3 1.41e3 1.47e3 1.32e3 1.38e3

S3a 2.51e3 1.26e3 1.37e3 1.12e3 1.22e3

S04 1.44e3 4.85e2 5.10e2 4.18e2 4.75e2

S04 1.21e3 4.64e2 4.40e2 3.53e2 4.24e2

S04 1.39e3 4.90e2 4.77e2 4.21e2 4.98e2

S04 1.43e3 4.99e2 4.94e2 4.30e2 5.02e2

S04 1.25e3 4.24e2 4.17e2 3.73e2 4.11e2

S04 1.47e3 5.07e2 4.64e2 4.14e2 4.56e2

S04 1.06e3 4.01e2 4.00e2 3.53e2 3.58e2

S04 1.08e3 3.90e2 3.75e2 3.33e2 4.02e2

S04 1.23e3 4.58e2 3.94e2 3.53e2 4.13e2

S04 1.43e3 5.36e2 5.57e2 4.67e2 5.18e2

120 S04 1.36e3 4.76e2 4.82e2 4.14e2 4.27e2

S04 1.56e3 5.15e2 4.96e2 4.06e2 4.40e2

S04 1.64e3 5.39e2 5.39e2 4.37e2 4.67e2

S04 1.59e3 5.16e2 4.67e2 4.31e2 4.69e2

S04 1.66e3 5.51e2 5.30e2 4.32e2 4.86e2

S04 1.54e3 5.05e2 5.12e2 4.44e2 4.80e2

S04 1.50e3 4.75e2 4.38e2 4.08e2 4.14e2

S04 1.62e3 5.47e2 5.06e2 3.96e2 4.57e2

S05 3.89e2 1.42e2 1.18e2 1.12e2 1.66e2

S05 4.26e2 1.52e2 1.23e2 1.11e2 1.61e2

S05 4.40e2 1.51e2 1.23e2 1.12e2 1.79e2

S05 4.09e2 1.49e2 1.28e2 1.24e2 1.82e2

S05 4.68e2 1.58e2 1.33e2 1.22e2 1.83e2

S05 4.22e2 1.53e2 1.30e2 1.20e2 1.72e2

S05 2.72e2 9.71e1 8.64e1 7.23e1 1.12e2

S05 3.60e2 1.15e2 8.76e1 7.63e1 1.08e2

S05 3.10e2 1.00e2 7.72e1 1.05e2 1.46e2

S05 3.46e2 1.22e2 9.89e1 9.58e1 1.51e2

S05 3.36e2 1.19e2 9.07e1 9.21e1 1.48e2

S05 3.26e2 1.14e2 9.66e1 9.02e1 1.40e2

S05 3.25e2 1.20e2 8.34e1 7.70e1 1.18e2

S05 3.26e2 1.15e2 7.57e1 7.57e1 1.25e2

S05 3.01e2 1.17e2 7.65e1 7.69e1 1.37e2

121 S05 3.18e2 9.93e1 8.30e1 7.98e1 1.25e2

S05 3.02e2 1.08e2 8.80e1 7.95e1 1.20e2

S05 2.87e2 1.20e2 9.80e1 7.85e1 1.36e2

S06 5.44e2 2.05e2 1.90e2 1.96e2 2.25e2

S06 7.14e2 2.23e2 2.37e2 2.17e2 2.55e2

S06 5.41e2 2.09e2 1.85e2 1.87e2 2.31e2

S06 5.60e2 2.06e2 2.03e2 2.06e2 2.21e2

S06 5.38e2 1.97e2 1.83e2 1.82e2 2.13e2

S06 6.27e2 2.01e2 1.92e2 1.94e2 2.30e2

S06 6.81e2 2.45e2 2.36e2 2.29e2 2.37e2

S06 6.22e2 2.33e2 2.56e2 2.28e2 2.53e2

S06 8.14e2 2.99e2 2.94e2 2.84e2 3.01e2

S06 5.19e2 1.87e2 1.76e2 1.58e2 1.76e2

S06 4.69e2 1.67e2 1.70e2 1.45e2 1.62e2

S06 5.32e2 1.98e2 1.83e2 1.61e2 1.87e2

S06 6.38e2 2.40e2 2.60e2 2.27e2 2.69e2

S06 7.12e2 2.79e2 2.71e2 2.48e2 2.70e2

S06 5.74e2 2.12e2 2.09e2 1.89e2 2.13e2

S07 1.19e3 5.34e2 4.11e2 4.64e2 4.69e2

S07 1.30e3 5.33e2 4.81e2 4.29e2 5.73e2

S07 1.33e3 5.96e2 5.19e2 4.40e2 5.01e2

S07 1.42e3 6.33e2 5.95e2 6.41e2 6.34e2

S07 1.47e3 6.19e2 5.82e2 4.90e2 5.36e2

122 S07 1.45e3 5.64e2 5.58e2 5.27e2 6.22e2

S07 1.32e3 5.47e2 5.77e2 5.38e2 7.17e2

S07 1.46e3 6.22e2 5.98e2 6.34e2 6.65e2

S07 1.44e3 5.77e2 5.37e2 4.59e2 5.52e2

S07 9.57e2 2.92e2 2.30e2 1.73e2 2.32e2

S07 9.24e2 3.03e2 2.27e2 1.86e2 2.66e2

S07 9.48e2 2.74e2 2.34e2 1.61e2 3.50e2

S07 1.05e3 3.20e2 2.30e2 1.82e2 2.84e2

S07 8.23e2 2.63e2 3.37e2 2.17e2 2.80e2

S07 1.04e3 3.52e2 2.29e2 1.57e2 2.53e2

S07 8.55e2 2.78e2 2.12e2 1.93e2 2.98e2

S07 9.48e2 3.08e2 2.41e2 1.72e2 3.35e2

S07 9.69e2 3.25e2 2.68e2 1.63e2 2.61e2

S08 2.45e3 1.13e3 1.06e3 9.73e2 1.03e3

S08 2.45e3 1.18e3 1.05e3 1.02e3 1.03e3

S08 2.57e3 1.17e3 1.06e3 1.01e3 1.02e3

S08 2.17e3 1.21e3 1.19e3 1.13e3 1.12e3

S08 2.00e3 1.18e3 1.12e3 1.11e3 1.15e3

S08 2.32e3 1.22e3 1.16e3 1.11e3 1.15e3

S08 2.37e3 9.79e2 9.07e2 8.28e2 8.47e2

S08 2.33e3 9.93e2 9.15e2 8.72e2 8.95e2

S08 2.23e3 9.48e2 9.09e2 8.41e2 8.89e2

S08 2.32e3 9.91e2 8.74e2 8.54e2 8.58e2

123 S08 2.40e3 9.99e2 9.40e2 8.53e2 8.39e2

S08 2.42e3 1.04e3 8.97e2 8.41e2 8.39e2

S08 1.57e3 7.36e2 7.21e2 6.61e2 6.46e2

S08 1.64e3 8.23e2 7.29e2 6.79e2 6.79e2

S08 1.56e3 7.64e2 7.96e2 6.96e2 6.82e2

S08 1.44e3 6.54e2 5.46e2 4.55e2 4.94e2

S08 1.41e3 6.08e2 5.46e2 4.70e2 4.67e2

S08 1.37e3 6.29e2 5.01e2 4.67e2 4.68e2

S06 5.34e2 2.22e2 2.04e2 1.88e2 2.35e2

S06 5.20e2 2.18e2 2.03e2 1.88e2 2.30e2

S06 5.28e2 2.11e2 1.98e2 1.85e2 2.22e2

S06 4.50e2 1.35e2 1.23e2 1.18e2 1.49e2

S06 5.06e2 1.45e2 1.11e2 1.22e2 1.44e2

S06 5.64e2 1.61e2 1.25e2 1.18e2 1.44e2

S06 5.73e2 2.37e2 2.06e2 2.28e2 2.86e2

S06 5.63e2 2.36e2 2.37e2 2.32e2 2.82e2

S06 5.63e2 2.37e2 2.30e2 2.31e2 2.87e2

S06 5.93e2 2.42e2 1.96e2 2.06e2 2.61e2

S06 6.47e2 2.38e2 2.08e2 2.16e2 2.60e2

S06 3.75e2 1.51e2 1.55e2 1.90e2 2.11e2

S10 2.79e3 1.42e3 1.58e3 1.40e3 1.43e3

S10 2.74e3 1.40e3 1.57e3 1.38e3 1.39e3

S10 2.96e3 1.62e3 1.65e3 1.54e3 1.60e3

124 S10 2.70e3 1.51e3 1.55e3 1.39e3 1.40e3

S10 2.49e3 1.33e3 1.41e3 1.34e3 1.31e3

S10 2.44e3 1.28e3 1.40e3 1.26e3 1.27e3

S10 2.99e3 1.46e3 1.57e3 1.44e3 1.50e3

S10 2.46e3 1.28e3 1.38e3 1.23e3 1.23e3

S10 2.64e3 1.31e3 1.44e3 1.25e3 1.34e3

S10 3.22e3 1.63e3 1.87e3 1.62e3 1.58e3

S10 3.07e3 1.63e3 1.68e3 1.51e3 1.56e3

S10 3.02e3 1.58e3 1.75e3 1.54e3 1.49e3

S10 2.28e3 1.16e3 1.28e3 1.12e3 1.19e3

S10 2.63e3 1.43e3 1.52e3 1.42e3 1.48e3

S10 2.43e3 1.32e3 1.51e3 1.42e3 1.42e3

S11 4.24e2 5.12e1 3.73e1 4.45e1 1.18e2

S11 4.74e2 8.09e1 3.32e1 3.93e1 9.86e1

S11 4.56e2 7.35e1 3.34e1 4.17e1 1.04e2

S11 3.96e2 7.81e1 3.58e1 4.39e1 9.40e1

S11 3.98e2 5.98e1 3.03e1 4.38e1 1.13e2

S11 3.40e2 6.26e1 3.61e1 4.63e1 1.19e2

S11 4.58e2 7.86e1 4.27e1 4.13e1 1.10e2

S11 5.12e2 1.05e2 5.06e1 5.49e1 1.42e2

S11 4.93e2 9.87e1 5.69e1 4.92e1 1.40e2

S11 5.03e2 7.67e1 3.16e1 2.99e1 1.23e2

S11 4.73e2 7.99e1 4.48e1 3.26e1 1.06e2

125 S11 4.64e2 8.85e1 4.00e1 4.42e1 1.12e2

S11 5.29e2 8.56e1 4.36e1 4.36e1 1.17e2

S11 5.17e2 9.24e1 3.78e1 3.84e1 1.15e2

S11 4.53e2 9.76e1 3.84e1 3.93e1 1.03e2

S11 4.26e2 9.91e1 4.79e1 3.99e1 1.10e2

S11 3.99e2 8.77e1 4.14e1 3.82e1 1.06e2

S11 3.83e2 8.34e1 3.50e1 3.28e1 9.43e1

S12 1.08e3 4.12e2 3.74e2 3.52e2 4.09e2

S12 8.59e2 3.27e2 3.02e2 2.74e2 3.71e2

S12 1.03e3 3.51e2 3.22e2 3.64e2 3.92e2

S12 7.12e2 2.76e2 2.98e2 2.56e2 3.75e2

S12 6.65e2 2.56e2 2.36e2 2.77e2 2.81e2

S12 6.88e2 2.47e2 2.86e2 2.50e2 3.27e2

S12 6.88e2 2.47e2 2.86e2 2.50e2 3.27e2

S12 7.13e2 2.56e2 2.69e2 2.48e2 2.94e2

S12 3.75e2 1.47e2 1.48e2 2.00e2 2.34e2

S12 8.11e2 2.78e2 2.42e2 2.28e2 2.87e2

S12 7.20e2 2.68e2 2.28e2 2.33e2 2.85e2

S12 7.32e2 2.70e2 2.29e2 3.03e2 2.97e2

S12 1.29e3 3.63e2 4.31e2 4.23e2 4.75e2

S12 1.21e3 3.63e2 4.74e2 4.86e2 5.70e2

S12 1.35e3 3.32e2 4.00e2 4.08e2 5.26e2

S12 7.30e2 2.18e2 1.99e2 1.94e2 2.62e2

126 S12 7.20e2 2.39e2 2.00e2 2.08e2 2.57e2

S12 3.32e2 1.05e2 1.18e2 1.44e2 1.57e2

S13 1.22e3 4.93e2 5.05e2 4.94e2 5.40e2

S13 1.29e3 5.53e2 5.45e2 5.18e2 5.58e2

S13 1.37e3 5.63e2 5.33e2 5.50e2 5.52e2

S13 1.45e3 5.79e2 5.52e2 5.41e2 5.80e2

S13 1.69e3 6.31e2 6.24e2 5.63e2 6.41e2

S13 1.72e3 6.08e2 5.96e2 5.73e2 6.06e2

S13 1.15e3 4.58e2 4.24e2 4.03e2 3.65e2

S13 1.20e3 4.27e2 3.99e2 4.20e2 4.23e2

S13 1.21e3 4.70e2 4.18e2 3.94e2 4.12e2

S13 1.48e3 6.35e2 6.56e2 5.95e2 6.02e2

S13 1.62e3 6.79e2 6.59e2 6.06e2 6.39e2

S13 1.80e3 6.87e2 6.47e2 6.44e2 6.62e2

S13 1.55e3 6.93e2 7.09e2 6.64e2 7.00e2

S13 1.78e3 7.91e2 7.27e2 7.04e2 7.37e2

S13 1.76e3 7.25e2 7.10e2 6.89e2 7.29e2

S13 1.02e3 3.92e2 3.62e2 3.31e2 3.65e2

S13 1.03e3 4.04e2 4.11e2 3.69e2 3.59e2

S13 1.06e3 4.26e2 3.79e2 3.83e2 3.79e2

S14 2.50e3 1.13e3 1.10e3 1.03e3 9.94e2

S14 2.66e3 1.14e3 1.12e3 1.04e3 1.06e3

S14 2.71e3 1.16e3 1.18e3 1.06e3 1.14e3

127 S14 4.36e3 1.97e3 1.86e3 1.78e3 1.81e3

S14 4.14e3 1.78e3 1.88e3 1.79e3 1.86e3

S14 4.69e3 1.87e3 1.85e3 1.80e3 1.87e3

S14 2.90e3 1.40e3 1.31e3 1.34e3 1.38e3

S14 3.71e3 1.45e3 1.41e3 1.37e3 1.47e3

S14 3.35e3 1.34e3 1.39e3 1.39e3 1.36e3

S14 3.27e3 1.40e3 1.33e3 1.31e3 1.28e3

S14 3.20e3 1.35e3 1.32e3 1.23e3 1.24e3

S14 3.50e3 1.43e3 1.43e3 1.32e3 1.27e3

S14 2.03e3 7.76e2 8.32e2 7.39e2 7.05e2

S14 1.97e3 7.75e2 7.81e2 7.63e2 7.26e2

S14 2.06e3 7.80e2 7.69e2 7.16e2 7.22e2

S14 3.02e3 1.25e3 1.21e3 1.07e3 1.10e3

S14 2.71e3 1.25e3 1.16e3 1.12e3 1.14e3

S14 2.92e3 1.21e3 1.21e3 1.19e3 1.14e3

S07 1.31e3 4.85e2 4.76e2 3.83e2 4.25e2

S07 1.31e3 4.40e2 4.83e2 3.88e2 4.49e2

S07 1.16e3 4.53e2 4.38e2 3.43e2 4.11e2

S07 1.53e3 5.56e2 5.08e2 4.73e2 5.20e2

S07 1.34e3 5.10e2 5.19e2 4.19e2 4.50e2

S07 1.53e3 5.40e2 5.32e2 4.59e2 5.22e2

S07 1.81e3 6.08e2 5.97e2 4.72e2 5.24e2

S07 1.80e3 5.73e2 5.40e2 4.69e2 5.26e2

128 S07 1.61e3 5.58e2 5.53e2 4.80e2 5.17e2

S07 1.22e3 3.86e2 3.76e2 3.37e2 3.45e2

S07 1.24e3 4.09e2 3.93e2 3.45e2 3.59e2

S07 1.15e3 3.89e2 4.16e2 3.13e2 3.30e2

S07 1.47e3 5.12e2 4.67e2 4.00e2 4.27e2

S07 1.58e3 5.17e2 4.77e2 4.43e2 4.80e2

S07 1.55e3 5.31e2 5.03e2 4.11e2 4.58e2

S07 1.70e3 5.03e2 4.75e2 4.24e2 4.58e2

S07 1.57e3 4.57e2 4.42e2 3.95e2 4.36e2

S07 1.69e3 4.85e2 4.74e2 4.08e2 4.26e2

; run;

Proc GLM Data= Frisch_20080215_Method1;

Class Sample;

Model DP23 DP4 DP5 DP6 DP7 = Sample;

LSMeans Sample / PDIFF ADJUST=Tukey;

LSMeans Sample;

Manova H = Sample / MSTAT = exact; run;

129 APPENDIX B: SAS CODE FOR DENIM SAMPLES USING TOTAL MOLES OF MONOSACCHARIDES

130 Data Frisch_20080226_Method2;

Input #1 @1 Sample $3.

@5 DP1 7.;

Datalines;

S1a 9.66e-8

S1a 9.66e-8

S1a 9.79e-8

S1a 6.22e-8

S1a 5.66e-8

S1a 6.35e-8

S1a 7.72e-8

S1a 8.03e-8

S1a 8.64e-8

S1a 6.22e-8

S1a 5.95e-8

S1a 6.39e-8

S1a 9.08e-8

S1a 9.06e-8

S1a 9.04e-8

S1a 9.13e-8

S1a 9.25e-8

S1a 9.06e-8

S02 1.42e-7

131 S02 1.28e-7

S02 1.33e-7

S02 1.37e-7

S02 1.23e-7

S02 1.33e-7

S02 1.68e-7

S02 1.60e-7

S02 1.46e-7

S02 1.48e-7

S02 1.28e-7

S02 1.39e-7

S02 1.37e-7

S02 1.40e-7

S02 1.59e-7

S02 1.62e-7

S02 1.28e-7

S02 1.39e-7

S3a 4.42e-8

S3a 4.45e-8

S3a 4.71e-8

S3a 5.59e-8

S3a 5.27e-8

S3a 5.16e-8

132 S3a 6.43e-8

S3a 6.02e-8

S3a 5.65e-8

S3a 5.37e-8

S3a 4.92e-8

S3a 4.90e-8

S3a 5.27e-8

S3a 4.79e-8

S3a 4.49e-8

S04 1.99e-8

S04 1.73e-8

S04 1.96e-8

S04 2.01e-8

S04 1.72e-8

S04 1.98e-8

S04 1.54e-8

S04 1.54e-8

S04 1.71e-8

S04 2.10e-8

S04 1.89e-8

S04 2.05e-8

S04 2.17e-8

S04 2.08e-8

133 S04 2.19e-8

S04 2.08e-8

S04 1.94e-8

S04 2.11e-8

S05 5.55e-9

S05 5.83e-9

S05 6.01e-9

S05 5.95e-9

S05 6.37e-9

S05 5.97e-9

S05 3.83e-9

S05 4.46e-9

S05 4.43e-9

S05 4.87e-9

S05 4.70e-9

S05 4.59e-9

S05 4.33e-9

S05 4.29e-9

S05 4.24e-9

S05 4.22e-9

S05 4.18e-9

S05 4.31e-9

S06 8.15e-9

134 S06 9.86e-9

S06 8.10e-9

S06 8.37e-9

S06 7.87e-9

S06 8.64e-9

S06 9.75e-9

S06 9.54e-9

S06 1.19e-8

S06 7.28e-9

S06 6.66e-9

S06 7.55e-9

S06 9.79e-9

S06 1.07e-8

S06 8.36e-9

S07 1.84e-8

S07 1.99e-8

S07 2.03e-8

S07 2.35e-8

S07 2.22e-8

S07 2.23e-8

S07 2.22e-8

S07 2.38e-8

S07 2.14e-8

135 S07 1.13e-8

S07 1.14e-8

S07 1.18e-8

S07 1.23e-8

S07 1.15e-8

S07 1.22e-8

S07 1.10e-8

S07 1.20e-8

S07 1.19e-8

S08 3.98e-8

S08 4.04e-8

S08 4.09e-8

S08 4.10e-8

S08 3.93e-8

S08 4.17e-8

S08 3.56e-8

S08 3.60e-8

S08 3.49e-8

S08 3.53e-8

S08 3.61e-8

S08 3.61e-8

S08 2.60e-8

S08 2.73e-8

136 S08 2.70e-8

S08 2.15e-8

S08 2.10e-8

S08 2.06e-8

S06 8.29e-9

S06 8.15e-9

S06 8.05e-9

S06 5.84e-9

S06 6.14e-9

S06 6.65e-9

S06 9.17e-9

S06 9.30e-9

S06 9.28e-9

S06 8.98e-9

S06 9.40e-9

S06 6.49e-9

S10 5.17e-8

S10 5.08e-8

S10 5.62e-8

S10 5.13e-8

S10 4.73e-8

S10 4.59e-8

S10 5.38e-8

137 S10 4.55e-8

S10 4.79e-8

S10 5.95e-8

S10 5.67e-8

S10 5.62e-8

S10 4.22e-8

S10 5.09e-8

S10 4.86e-8

S11 4.02e-9

S11 4.33e-9

S11 4.22e-9

S11 3.86e-9

S11 3.85e-9

S11 3.61e-9

S11 4.35e-9

S11 5.16e-9

S11 5.00e-9

S11 4.69e-9

S11 4.55e-9

S11 4.46e-9

S11 4.81e-9

S11 4.85e-9

S11 4.33e-9

138 S11 4.00e-9

S11 3.94e-9

S11 3.86e-9

S12 1.58e-8

S12 1.28e-8

S12 1.47e-8

S12 1.15e-8

S12 1.03e-8

S12 1.08e-8

S12 1.08e-8

S12 1.07e-8

S12 6.63e-9

S12 1.10e-8

S12 1.04e-8

S12 1.10e-8

S12 1.79e-8

S12 1.86e-8

S12 1.81e-8

S12 9.60e-9

S12 9.72e-9

S12 5.13e-9

S13 1.95e-8

S13 2.08e-8

139 S13 2.14e-8

S13 2.22e-8

S13 2.48e-8

S13 2.46e-8

S13 1.68e-8

S13 1.71e-8

S13 1.74e-8

S13 2.38e-8

S13 2.52e-8

S13 2.66e-8

S13 2.59e-8

S13 2.84e-8

S13 2.77e-8

S13 1.48e-8

S13 1.54e-8

S13 1.57e-8

S14 4.05e-8

S14 4.21e-8

S14 4.35e-8

S14 7.07e-8

S14 6.87e-8

S14 7.24e-8

S14 5.00e-8

140 S14 5.64e-8

S14 5.29e-8

S14 5.15e-8

S14 5.00e-8

S14 5.36e-8

S14 3.04e-8

S14 3.00e-8

S14 3.03e-8

S14 4.58e-8

S14 4.43e-8

S14 4.60e-8

S07 1.84e-8

S07 1.84e-8

S07 1.68e-8

S07 2.15e-8

S07 1.94e-8

S07 2.14e-8

S07 2.40e-8

S07 2.34e-8

S07 2.22e-8

S07 1.59e-8

S07 1.64e-8

S07 1.56e-8

141 S07 1.96e-8

S07 2.09e-8

S07 2.06e-8

S07 2.13e-8

S07 1.98e-8

S07 2.08e-8

; run;

Proc GLM Data= Frisch_20080226_Method2;

Class Sample;

Model DP1 = Sample;

LSMeans Sample / PDIFF ADJUST=Tukey;

LSMeans Sample; run;

142 LIST OF REFERENCES

1. Qureshi, S.N., et al., Molecular Biology and Physiology EST-SSR: A New Class of Genetic Markers in Cotton. Journal of Cotton Science, 2004. 8: p. 112-123.

2. Mathews, C.K., K.E. van Holde, and K.G. Ahern, Biochemistry. Third ed. 2000: Robin Heyden.

3. Gaudette, B.D., The Forensic Aspects of Textile Fiber Examination, in Forensic Science Handbook, R. Saferstein, Editor. 1988, Prentice Hall: Englewood Cliffs, NJ. p. 209-272.

4. Wakelyn, P.J., et al., International Fiber Science and Technology: Cotton Fiber Chemistry and Technology, in Previously published in the Handbook of Fiber Chemistry, Third Edition, Taylor & Francis, 2007., M.L.I.F.S.a.T.S. Editor), Editor. 2007, CRC Press, Taylor & Francis Group, LL.

5. Price, J.B., et al., Assessing the Quality of Four Naturally Colored Cottons. Textile Research Journal, 2001. 71(11): p. 993-999.

6. Ryser, U., Cell Wall Biosynthesis in Differentiating Cotton Fibres. European Journal of Cell Biology, 1985. 39: p. 236-256.

7. Pfluger, J. and P.C. Zambryski, Cell Growth: The Power of Symplastic Isolation. Current Biology, 2001. 11(11): p. R436-R439.

8. Murray, A.K. and J. Brown. Glycoconjugate Analysis of Developing Cotton Fibers from Several Varieties Grown on the Same Site. in Proceedings - Beltwide Cotton Conferences. 1996.

9. Murray, A.K., Using Glycan Oligomer Analysis for Examination of Textile Fibers, Paper, Wood, Grains, Food and Other Cellulose Containing Materials of Plant Origin, in http://www.freepatentsonline.com, W.I.P. Organization, Editor. 2002: United States. p. 1-57.

10. Wood, T.M., Cellulose and Cellulolysis. World Review of Nutrition and Dietetics, 1970. 12: p. 227-265.

11. Shenouda, S.G., Chapter 7. The Structure of Cotton Cellulose, in Applied Fiber Science, F. Happey, Editor. 1979, Academic: London. p. 275-309.

12. Zhang, Y.Z., et al., Size and Arrangement of Elementary Fibrils in Crystalline Cellulose Studied with Scanning Tunneling Microscopy. Journal of Vacuum Science Technology B. - Microelectronics and Nanometer Structures, 1997. 15(4): p. 1502-1505.

143

13. Delmer, D.P. and Y. Amor, Cellulose Biosynthesis. The Plant Cell, 1995. 7: p. 987-1000.

14. Saxena, I.M. and R.M.B.J. Brown, Cellulose Biosynthesis: Current Views and Evolving Concepts. Annals of Botany, 2005. 96: p. 9-21.

15. Brett, C.T., Cellulose Microfibrils in Plants: Biosynthesis, Deposition, and Integration into the Cell Wall. International Review of Cytology, 2000. 199: p. 161-1999.

16. Gilbert, R., Chain Modulus and Intramolecular Hydrogen Bonding Native and Regenerated Cellulose Fibers. Chemtracts-Macromolecular Chemistry, 1992. 3: p. 142- 144.

17. United States Department of Agriculture, A.M.S., Cotton Classification: Understanding the Data, in Cotton Program. 2004, United States Department of Agriculture. p. 13.

18. Barnes, E., Email: FW: Contact Cotton Incorporated: North America, J. Frisch, Editor. 2008: Orlando.

19. Cotton Incorporated, ISP 1010: Denim Fabric Manufacturing, in Technical Bulletin. 2004, Cotton Incorporated: Cary, NC. p. 22.

20. Grieve, M.C., J. Dunlop, and P. Haddock, An Assessment of the Value of Blue, Red, and Black Cotton Fibers as Target Fibers in Forensic Science Investigations. Journal of Forensic Sciences, 1988. 33(6): p. 1332-1344.

21. Senior, J., R.A. Guest, and W.E. Wood, Sulfur Dyes, in Kirk-Othmer Encyclopedia of Chemical Technology. 1997, John Wiley & Sons, Inc. p. 169-189.

22. Rendle, D.F., Advances In Chemistry Applied To Forensic Science. Chemical Society Reviews, 2005. 34: p. 1021-1030.

23. Trace Evidence Analysis: More Cases In Mute Witnesses, ed. M.M. Houck. 2004: Elsevier Academic Press.

24. Cantrell, S., et al., A Textile Fibre Survey As An Aid To The Interpretation Of Fibre Evidence In The Sydney Region. Forensic Science International, 2001. 123: p. 48-53.

25. Palmer, R. and S. Oliver, The Population of Coloured Fibres in Human Head Hair. Science & Justice, 2004. 44(2): p. 83-88.

26. Watt, R., C. Roux, and J. Robertson, The Population of Coloured Textile Fibres in Domestic Washing Machines. Science & Justice, 2005. 45(2): p. 75-83.

144 27. Houck, M.M., Inter-comparison of Unrelated Fiber Evidence. Forensic Science International, 2003. 135: p. 146-149.

28. Wiggins, K.G., The European Fibres Group (EFG) 1193-2002: "Understanding and Improving the Evidential Value of Fibres". Analytical Bioanalytical Chemistry, 2003. 376: p. 1172-1177.

29. Murray, A.K., R.L. Nicholas, and G.F. Sassenrath-Cole, Cell wall biosynthesis: glycan containing oligomers in developing cotton fibers, cotton fabric, wood and paper. Phytochemistry, 2001. 57: p. 975-986.

30. Murray, A.K., Enzymatic Degradation of Cotton Fibres: Effect of Protein Cross-Linking and the Use of Degradation to Characterize Fibers of Plant of Different Genetic Background, in Patent Cooperation Treaty (PCT), W.I.P. Organization, Editor. 2003: United States. p. 1-47.

31. Murray, A.K. Characterization of cotton cultivars by glycan oligomer analysis. in Beltwide Cotton Conferences. 2006. San Antonio, Texas.

32. Murray, A.K. Glycan oligomer profiles of cotton. in Beltwide Cotton Conferences. 2006. San Antonio, Texas.

33. Murray, A.K., Method for monitoring textile fiber quality, and for analysis and identification of paper, wood and other cellulose containing materials. 2003, Alexander; Lyle A.: U.S.

34. Liu, Y., et al., Separation and characterization of underivatized oligosaccharides using liquid chromatography and liquid chromatography-electrospray ionization mass spectrometry. Journal of Chromatography A, 2005. 1079: p. 146-152.

35. Mauri, P., et al., Analysis of Saccharides in Beer Samples by Flow Injection with Electrospray Mass Spectrometry. Rapid Communications in Mass Spectrometry, 2002. 16: p. 743-748.

36. Wan, E.C.H. and J.Z. Yu, Determination of sugar compounds in atmospheric aerosols by liquid chromatography combined with positive electrospray ionization mass spectrometry. Journal of Chromatography, A, 2006. 1107: p. 175-181.

37. Choi, S.-S. and S.-H. Ha, Characterization of Ionized Maltooligosaccharides by Sodium Cation in MALDI-TOFMS Depending on the Molecular Size. Bull. Korean Chem. Soc., 2006. 27(8): p. 1243-1245.

38. Araujo, A.S., et al., Electrospray Ionization Mass Spectrometry Fingerprinting of Beer. Analyst, 2005. 130: p. 884-889.

145 39. Redmond, J.W. and N.H. Packer, The Use of Solid-Phase Extraction with Graphitised Carbon for the Fractionation and Purification of Sugars. Carbohydrate Research, 1999. 319: p. 74-79.

40. Packer, N.H., et al., A general approach to desalting oligosaccharides released from glycoproteins. Glycoconjugate Journal, 1998. 15: p. 737-747.

41. Suzuki, S., K. Kakehi, and S. Honda, Comparison of the Sensitivities of Various Derivatives of Oligosaccharides in LC/MS with Fast Atom Bombardment and Electrospray Ionization Interfaces. Analytical Chemistry, 1996. 68: p. 2073-2083.

42. Li, M. and J.A. Kinzer, Structural Analysis of Oligosaccharides by a Combination of Electrospray Mass Spectrometry and Bromine Isotope Tagging of Reducing-End Sugars with 2-Amino-5-Bromophyridine. Rapid Communications in Mass Spectrometry, 2003. 17: p. 1462-1466.

43. Zaia, J., Mass Spectrometry of Oligosaccharides. Mass Spectrometry Reviews, 2004. 23: p. 161-227.

44. Gilbert, C. and S. Kokot, Discrimination of cellulosic fabrics by diffuse reflectance infrared Fourier transform spectroscopy and chemometrics. Vibrational Spectroscopy, 1995. 9: p. 161-167.

45. van der Peijl, G.J.Q. (GC-)IRMS Applications for Jeans and Motor Oil Investigations. in American Academy of Forensic Sciences. 2007. San Antonio, Texas.

46. Martsinkovskaya, A.I., R.S. Moukhamedov, and A.A. Abdukarimov, Potential Use of PCR-Amplified Ribosomal Intergenic Sequences for Differentiation of Varieties and Species of Gossypium Cotton. Plant Molecular Biology Reporter, 1996. 14(1): p. 44-49.

47. Suzuki, S., et al., Microspectrophotometric discrimination of single fibers dyed by indigo and its derivatives using ultraviolet-visible transmittance spectra. Science & Justice, 2001. 41(2): p. 107-111.

48. Herbert, C.G. and R.A.W. Johnstone, Mass Spectrometry Basics. 2003: CRC Press LLC. 474.

49. Electrospray Ionization Mass Spectrometry: Fundamentals, Instrumentation & Applications, ed. R.B. Cole. 1997: John Wiley & Sons, Inc.

50. Gabelica, V. and E.D. Pauw, Internal Energy and Fragmentation of Ions Produced In Electrospray Sources. Mass Spectrometry Reviews, 2005. 24: p. 566-587.

146 51. Nguyen, S. and J.B. Fenn. Gas-phase ions of solute species from charged droplets of solutions. in Proceedings of the National Academy of Sciences of the United States of America. 2007: National Academy of Sciences.

52. Tang, K., J.S. Page, and R.D. Smith, Charge competition and the linear dynamic range of detection in electrospray ionization mass spectrometry. American Society for Mass Spectrometry, 2004. 15: p. 1416-1423.

53. March, R.E., An Introduction to Quadrupole Ion Trap Mass Spectrometry. Journal of Mass Spectrometry, 1997. 32: p. 351-369.

54. Wong, P.S.H. and R.G. Cooks, Ion trap mass spectrometry. Bioanalytical Systems, Inc., 1997. 16(3): p. 85-92.

55. Lam, H., Performance Verification of HPLC. Analytical Method Validation and Instrument Performance Verification, ed. H.L. Chung Chow Chan, Y.C. Lee, and Xue- Ming Zhang. 2004: John Wiley & Sons, Inc.

56. O'Rourke, N., L. Hatcher, and E.J. Stepanski, A Step-by-Step Approach to Using SAS for Univariate & Multivariate Statistics. Second ed. 2005, Cary, NC: SAS Institute and John Wiley & Sons.

57. Healey, J.F., Statistics: A Tool for Social Research. 6th ed. 2002: Wadsworth Group. 519.

58. Lowry, R. Chapter 14: One-Way Analysis of Variance for Independent Samples, Part 2. 2008 [cited March 13, 2008]; Available from: http://faculty.vassar.edu/lowry/ch14pt2.html.

59. SYSTAT 11: Statistics II, ed. I. SYSTAT Software. 2004, Richmond, CA. II-505-II-508.

60. HyPURITY AQUASTAR Columns. 2007 December 27, 2007 [cited; Available from: http://www.thermo.com/com/cda/product/detail/0,1055,22219,00.html.

61. Limited, T.I.C.A., Table: HVI Data from Cotton Samples, N.C.F.S. Jessica Frisch, Editor. 2008: Orlando. p. Cotton Samples with Table of HVI Data.

62. Laing, D.K., et al., The Extraction and Classification of Dyes from Cotton and Viscose Fibres. Forensic Science International, 1991. 50: p. 23-35.

63. Reidies, A.H., D. Jensen, and M. Guisti, How oxidation of indigo contributes to the yellowing of acid washed denim. Textile Chemist and Colorist, 1992. 24(5): p. 26-30.

147